Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
i
ATTITUDINAL MODELING OF
AFFECT, BEHAVIOR AND COGNITION
Semantic Mining of Disaster Text Corpus
HALIMAHTUN M. KHALID, JENTHI K. RADHA
Damai Sciences Sdn Bhd, Malaysia
MARTIN G. HELANDER, JESSIE XI YANG
Nanyang Technological University, Singapore
Air Force Office of Scientific Research (AFOSR)
Asian Office of Aerospace Research and Development (AOARD)
Award Number FA2386‐09‐1‐4009
October 2010
Report Documentation Page Form ApprovedOMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering andmaintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, ArlingtonVA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if itdoes not display a currently valid OMB control number.
1. REPORT DATE 12 NOV 2010
2. REPORT TYPE Final
3. DATES COVERED 25-06-2009 to 22-10-2010
4. TITLE AND SUBTITLE Attitudinal Modeling of Affect, Behavior and Cognition
5a. CONTRACT NUMBER FA23860914009
5b. GRANT NUMBER
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) Halimahtun Khalid
5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) Damai Sciences Sdn Bhd,A-31-3 Suasana Sentral,Jalan Stesen Sentral5,Kuala Lumpur 50470 Malaysia,NA,NA
8. PERFORMING ORGANIZATIONREPORT NUMBER N/A
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) AOARD, UNIT 45002, APO, AP, 96338-5002
10. SPONSOR/MONITOR’S ACRONYM(S) AOARD
11. SPONSOR/MONITOR’S REPORT NUMBER(S) AOARD-094009
12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES
14. ABSTRACT This document presents an investigation into the attitudes of disaster experience from text corpora ofnarratives and reports. Attitude is made up of three components: Affect, Behavior, and Cognition (ABC).The narratives comprised personal stories obtained from Internet blogs, and the reports were research andmedia articles about people?s experiences derived from digital libraries. This study identifies the semanticsof disaster experience for development of hybrid ontologies that may be used to model attitudes in the faceof disasters. Different disaster scenarios revealed different mapping relationships of ABC.
15. SUBJECT TERMS sociocultural modeling
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF ABSTRACT Same as
Report (SAR)
18. NUMBEROF PAGES
112
19a. NAME OFRESPONSIBLE PERSON
a. REPORT unclassified
b. ABSTRACT unclassified
c. THIS PAGE unclassified
Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18
ii
Acknowledgments
This report is the result of a one year research investigation by Damai Sciences, Malaysia
and Nanyang Technological University, Singapore. The study was supported by a grant
from the Asian Office of Aerospace Research and Development (AOARD), Award No.
FA2386-09-1-4009, under the AFOSR Social Cultural Modeling Program.
We would like to express our appreciation to the many people who contributed to the
research:
Dr Terence Lyons of AFOSR for the grant support and feedback on the research;
Dr. Kenneth Goretta, as AOARD program manager, for research and travel support;
Dr Kenneth Boff for his critique of manuscripts;
Dr David Sonntag for establishing contacts with various researchers;
Dr. Andrew Smith for the use of Leximancer 3.5.
We are also appreciative of the effort by Nur Amalina Abd Aziz and Faridah Hanim
Abd Malek in preparing the data for analysis.
The views and opinions expressed in this report are of the authors. Responsibility for
the final content of this report rests entirely with the principal author.
Halimahtun M. Khalid, PhD, CPE
Damai Sciences, Malaysia
iii
Executive Summary
This report presents an investigation into the attitudes of disaster experience from text corpora of narratives and reports. Attitude is made up of three components: Affect, Behavior, and Cognition (ABC). The narratives comprised personal stories obtained from Internet blogs, and the reports were research and media articles about people’s experiences derived from digital libraries. The objective of the study was to identify the semantics of disaster experience for development of hybrid ontologies that may be used to model attitudes in the face of disasters. The different disaster scenarios revealed different mapping relationships of ABC.
Three disaster classes were studied: Natural Disasters comprising of tsunami, earthquake, hurricane/typhoon and flood; Human Induced Disasters comprising of terrorism, fire, industrial and transport accidents; and Pandemic Disasters comprising of swine flu, SARS, HIV/AIDS and tuberculosis. These disasters were selected on the basis of their significance, in particular social, economic and physical. Most of these disasters have impacted people in Asia.
The analysis involved text mining using two tools: Latent Semantic Analysis and Leximancer. The Analysis of Variance was performed on the frequency of concepts mined to determine the capability of each tool in semantic processing. Leximancer proved to be usable and produced semantic maps that were required for developing ontologies. Two types of ontologies were derived: Attitudinal ontology and Disaster Situational (organizational) ontology.
The ontologies were expressed at two levels: generic and specific. A generic ontology describes the semantics of a combined disaster class (e.g., natural disaster), while a specific ontology describes the semantics of a disaster type (e.g., tsunami for natural disaster). The ontologies were integrated to create a hybrid ontology for attitudinal modeling of disaster management.
This study departs from previous work on attitudinal research as it commenced with identification of ABC concepts using text mining of disaster corpora to develop a semantic framework for a baseline attitudinal model. The goal was to frame the central concepts within an application domain, disaster risks. The study also departs from other disaster research as the measurement of people’s attitude occurs after a hybrid ontology of attitude and disaster situations has been established. This helps in a better design of disaster management system as the ontologies may be used to integrate heterogenous information on disaster as well as reducing conflicts in semantic interoperability.
Conceptualization of the model was based on secondary data alone; therefore, the next step would be to gather primary human data to validate the forecasting capability of the model.
iv
List of Abbreviations and Acronyms
ABC Affect, Behavior, Cognition
AIDS Acquired Immune Deficiency Syndrome
AOARD Asian Office of Aerospace Research and Development
DAMAI Damai Sciences Sdn Bhd
df degrees of freedom
DM Disaster Management
HID Human Induced Disaster
HIV Human Immunodeficiency Virus
LSA Latent Semantic Analysis
LXM Leximancer
ND Natural Disaster
NTU Nanyang Technological University
PD Pandemic Disaster
PS Personal Stories
RA Research Articles
SARS Severe Acute Respiratory Syndrome
SD Standard Deviation
TB Tuberculosis
v
Contents
Executive Summary, ii Acknowledgment, iii List of Abbreviations and Acronyms, iv Contents, v – vii 1 INTRODUCTION 1
Project Overview, 2 Significance of the research, 4 Scientific innovations, 5
Organization of the Report, 5 2 CONCEPTUAL FRAMEWORK 6
Related Work, 7 Risk Perception and Risk Taking Behavior, 8 Affect, Behavior and Cognition, 9
Affect and cognition, 10 Affect and behavior, 13 Cognition and behavior, 13
Attitudinal Forecasting Model, 14 Human Factors in Disaster Management, 14
ABC conceptual model, 16 3 DISASTER TAXONOMY 19
Disaster Classification, 19 Natural Disasters, 21 Human Induced Disasters, 23 Pandemic Disasters, 25
Consequences of Disasters, 26 Scope of research on disaster, 28
vi
4 SEMANTIC FRAMEWORK 30
Semantic Maps for Visualizing Information, 30 Ontologies, 30
Ontologies approaches, 31 Ontologies architecture, 31
Generic ontologies, 33 Attitudinal ontology, 33 Disaster situational ontology, 34
Text mining, 35 Latent Semantic Analysis, 35 Leximancer analysis, 38 5 RESEARCH METHODOLOGY 40
Text Preparation, 40 Corpus sample, 40
Data Analyses, 45 Pilot study, 46
6 STUDY 1. COMPARISONS OF TOOLS IN SEMANTIC MINING 48
Research Design, 48 Aim and hypothesis, 48 Method, 48
Text corpus, 48 Data analysis, 49
Results, 49 Cohen’s Kappa for inter-classifier reliability, 49
ANOVA analysis, 51 Semantic mining of similar and dissimilar words, 52 Word correlation, 55 Concept mapping, 56 Semantic maps, 57
Discussion, 60 Conclusion, 60
7 STUDY 2. ATTITUDINAL SEMANTIC MINING 61
Research Design, 61 Aim and Hypothesis, 61 Method, 61
Corpus sample, 61 Procedure, 61
vii
Results and Discussion, 62 Fleiss Kappa inter-classifier reliability, 62 ABC association within disaster class, 62 ABC association within corpus type, 65 ABC semantics of natural disaster corpus, 68 ABC semantics of human induced disaster corpus, 70 ABC semantics of pandemic disaster class corpus, 72 Disaster situations semantics of natural disaster class corpus, 73 Disaster situations semantics of human induced disaster class corpus, 75 Disaster situations semantics of pandemic disaster class corpus, 76
Conclusion, 78 8 HYBRID ONTOLOGIES FOR ATTITUDINAL MODELING 79
Hybrid Ontologies, 79 Attitudinal ontologies, 80 Disaster situational (organizational) ontologies, 84 Summary, 87
9 SUMMARY AND FUTURE WORK 88
Summary of Findings, 88 Future Work, 91
Research design, 91 Risk attitudinal board, 91
REFERENCES 94 APPENDIXES 104
A Listing of articles for building text corpora B Sample of specimen recording form C Fleiss Kappa results for each disaster type D ABC semantic maps of each disaster type E Disaster situational semantic maps of each disaster type F Outline of study for Phase 2
1
1
Introduction
“The modeling of cognition and action by individuals and groups is quite possibly
the most difficult task humans have yet undertaken. Developments in this area are
still in their infancy.”
National Research Council, 1998, p. 8.
Various models that have emerged in the literature are useful for forecasting a range of outcomes,
rather than making precise predictions (Zacharias, MacMillan & Hemel, 2008). Most research on
prediction is concerned with people's ability to anticipate the occurrence of future external events
(e.g., "will the price of stock go up or down?") or their own behavior (e.g., "am I likely to get
divorced in the next 10 years?"). To name, but a few: decision making under uncertainty (e.g.,
Kahneman & Tversky, 2000; Gilovich, Griffin, & Kahneman, 2002; Nisbett & Ross, 1980), the
accuracy of people's predictions about their future behavior (e.g., Osberg & Shrauger, 1986;
Wilson & LaFleur, 1995), and the effects of temporal perspective on prediction (e.g., Trope &
Liberman, 2010). A crucial form of prediction, people's ability to forecast their own attitudes in
relation to their feelings (affect), intent (behavior) and belief (cognition) about events that may
induce hazard or risk such as disasters has not been entirely overlooked, but the investigations
have been in piecemeal fashion. The reason being, research in this area is still challenged by
rigorous methodologies and measures.
The aim of this project is to contribute to our understanding of attitudes, in the context of risk
associated with disasters. The proposed model is intended to identify important concepts that make
up attitude that may be useful for forecasting possible outcomes rather than making precise
predictions. Figure 1.1 illustrates a useful way for conceptualizing the components of attitude:
Affect, Behavior, and Cognition (ABC), following the tripartite attitudinal model of Breckler
(1984).
In Figure 1.1, Affect refers to an emotional response, a gut reaction, or sympathetic nervous
activity. It can be measured by monitoring physiological responses (e.g., heart rate, galvanic skin
response) or by collecting verbal reports of feelings or mood. Behavior includes overt actions,
2
behavioral intentions, and verbal statements regarding behavior. Cognition consists of beliefs,
knowledge structures, perceptual responses, and thoughts (Breckler, 1984). People’s stories of
their real life experiences provide a reliable source for understanding the interactions of ABC,
while written reports of verbal statements from survivors can enhance further analysis.
Figure 1.1 Tripartite model of attitude structure (Rosenberg & Hovland, 1960)
The relationship between the attitudinal components may be used to forecast people’s
behavior in disaster situations, gauged from the specific affect, behavior and cognition semantics
derived from text mining. These semantics represent people’s experiences in the face of disasters.
Other components necessary for forecasting such as group values, situation awareness and risk
assessment may not be revealed from the data. These aspects of the project may be investigated in
future work.
PROJECT OVERVIEW
Figure 1.2 is an overview of the project involving two phases:
1. Identifying the components of attitude. This activity involved two studies: the first
study was to identify a tool that has the ability to mine and visualize ABC concepts
for the purpose of conceptualizing a semantic framework. The second study used the
3
tool to further analyze the data to develop generic and specific ontologies.
Secondary data was used for this activity.
2. Validation of attitudinal model for behavioral forecasting. This activity has the
purpose of validating the ABC concepts in the field with people who may not have
experienced disaster in order to identify the role of culture in risk assessment. Other
measures that are crucial for forecasting behavior such as group values and norms,
ability to predict in situation awareness, and temporal perspective may also be
studied.
Figure 1.2 Overview of project
The investigation in Phase 1 was a two-pronged approach. The literature on psychology and
human factors were reviewed to identify concepts that relate to attitudinal components of ABC. In
addition past disaster research was reviewed to provide a taxonomy of disasters for the study. A
conceptual model was formulated following the tripartite model of Breckler (1984). To
conceptualize the model, secondary data on three classes of disasters were gathered from digital
libraries and Internet blogs. Text mining tools were used to mine semantics for development of
ontologies. The derived ontologies will be used in Phase 2 to further validate the model in the
context of socio-cultural of people who have not experienced disaster. This phase requires
development of a tool that can measure specific ABC to enable forecasting. The database may be
useful for disaster management.
4
Significance of the Research
Jung (1921) defined attitude as a "readiness of the psyche to act or react in a certain way"
(Jung, 1971). There are two forms of attitude: rational attitude, which is subdivided into “thinking”
and “feeling” psychological functions, each with its attitude. Irrational attitude subdivides into
“sensing” and “intuition” psychological functions, each with its attitude. Thus, there are attitudes
linked to thinking, feeling, sensation, and intuition (Jung, [1921] 1971: par. 691). Missing from
Jung’s definition is the concept of behavior. Later psychologists defined attitude as comprising of
three components: affect, behavior and cognition (Breckler, 1984; Solomon, 1996). Several
models emerged to explain the connection between the ABC concepts. The ACT-R architecture
(Anderson, 1983; 1996; Taatgen & Anderson, 2008) focuses on the relationship between the
processes of cognition and behavior; the cognitive appraisal theory of emotion (Scherer, Shorr, &
Johnstone, 2001) on the mapping of affect and cognition. To develop a comprehensive model of
ABC requires measurement beyond subjective methods. Objective methods such as using fMRI
technique confirm that affect is integrated with cognition in the output of behavior. The mapping
of affect, behavior and cognition may be explored using semantic mining.
This study was an attempt to identify the semantics of attitudinal components through the use
of text mining. It departs from previous text mining research (e.g., Landauer, Foltz, & Laham,
1998) as it used a relevant dataset (i.e. disaster articles) to research the problem, and with a
considerable size of text corpus. The purpose is to mine ABC concepts from specific literature to
enable the development of a baseline attitudinal ontology. For example, crisis workers assess crisis
using types of reactions: affective (feeling), behavioral (doing), and cognitive (thinking).
Assessment of the affective domain may predict the primary reaction, which could be anger, fear,
or stress. Behavioral reactions include approach, avoidance, or immobility (fight, flight, or freeze).
Cognitive reactions refer to the victim’s perception of the event. Individuals may perceive
transgressions of their rights being violated, threats of potential for harm, or an experience of
irretrievable loss. These perceptions may occur in any domain of life: physical, psychological,
social, environmental, values and beliefs.
This study, therefore, departs from previous studies on attitude in that it first identified ABC
concepts from people’s real life experiences to build a model that can be used in the future to
measure attitudes that can be used to forecast behavior.
5
Scientific Innovations
From the foregoing, we note several scientific innovations in the current research:
1. The attitudinal model encompassed the components of Affect, behavior and Cognition
(ABC). This is not new but the synergistic view of affect and cognition within the concept
of citarasa is a recent interpretation of “emotional intent” that has yet to be integrated
within the context of disaster experience (Khalid et al., 2010).
2. The development of attitudinal ontologies through text mining is novel.
3. The use of context-dependent word corpus for different disaster scenarios can be used to
identify the semantic space. While the required number of corpus may be limited, the
specimens and sources that contributed to the semantic space may be varied and many.
This is a new development in semantic research.
4. ABC was conceptualized through two-tiered phases of activities; first using secondary
data of people’s narratives and reports of experiences. This established the initial semantic
space, besides identifying ABC concepts for the second-level activity of primary data
gathering, which is planned for the future. Top-down and bottom-up approaches are
necessary for refining and validating attitudinal models.
ORGANIZATION OF THE REPORT
The report is organized into nine chapters. Chapter 1 introduces the research problem and a
conceptual model following Breckler (1984). The structure of the report is outlined as well.
Chapter 2 is a review of the literature that is relevant to the ABC concepts, in order to
contextualize the research and to conceptualize a model for the research. Chapter 3 describes the
semantic framework and proposes an ontological model, and also describes the text mining tools,
Latent Semantic Analysis (SA) and Leximancer (LXM). Chapter 4 outlines a taxonomic
classification of disasters as documented in the literature, and identifies the types of disasters being
investigated. Chapter 5 describes the methodology that guided the overall study. This methodology
is one of the contributions of the study, besides the ontology model. Chapter 6 presents the first
study aimed at comparing two text mining tools Latent Semantic Analysis (LSA) and Leximancer
(LXM) on their capability in semantic processing, while Chapter 7 describes the second study that
mined ABC and disaster situational factors. The results from this investigation were applied to the
ontologies and are discussed in Chapter 8. The final Chapter 9 discusses the main findings and
future work to be undertaken in the next phase.
6
2
Conceptual Framework
“The starting point of any theory of risk must be that everyone willingly takes risks”
Adams (1995, p. 16)
Most of us have imagined what it might be like to experience a tsunami, a fire or an infectious flu.
We have ideas about what we might do or fail to do and how it might feel. We have fears that we
admit to openly and ones that we never discuss. The word disaster connotes panic, hysterical
crowds, destruction and rescuers. But our disaster attitudes can be quite different from the ones we
expect to experience.
In the face of a major disaster event, many people tend to react with emotion and emotion-
laden decisions (Grandien et al., 2005; Lerner et al., 2003). The 2004 East Asian tsunami disaster
had a profound psychological impact on many countries, not only those that were directly hit by
the tsunami waves. Sweden had an unusually high number of tourists visiting East Asia at the time
of the disaster, resulting in over six hundred Swedes being killed or missing. A consequence of
this tragedy, and the media attention it received (Mann, 2007), many Swedes felt deeply involved
and saddened (Grandien et al., 2005). The feelings elicited by such an event may also have an
impact on everyday decisions. In a famous example, Johnson and Tversky (1983) found that
incidental affect (i.e., a mood state) induced by reading a newspaper article influenced subsequent
risk judgments. Previous research in judgment and decision making has shown that people tend to
rely on their affective reactions when making decisions (Lichtenstein & Slovic, 2006; Peters, 2006;
Slovic et al., 2002).
In a disaster situation, then, people travel through 3 phases of the “survival arc” (Ripley,
2009): denial (affect) which can take the form of delay that can be fatal. How long the delay lasts
depends on how we calculate risk. We are slow to recognize exceptions but this does not mean we
waste time during this delay. Given the time to think, people need information just like they need
shelter and water. Their brains lack the patterns they need to make a good decision, so they search
for better data. After denial, we move into deliberation (cognition). We know something is terribly
7
wrong, but we don’t know what to do about it. How do we decide? The first thing to understand is
that nothing is normal. We think and perceive differently. We all share a basic fear response. There
is also the effect of the group or crowd on our deliberation. Finally, we reach the third phase: the
decisive moment (behavior). We have accepted that we are in danger, we have deliberated our
options. Now we take action. Many – if not most – people tend to shut down entirely in a disaster.
They seem to lose all awareness. But their paralysis can be strategic.
RELATED WORK
A review of 160 disaster studies by Norris et al. (2002) assessed the severity of psychological
impairment among adults exposed to community trauma. The outcome variable was assessed by
multiple measures, including psychological problems (e.g., posttraumatic stress disorder [PTSD],
depression, and anxiety), non-specific distress, and health problems. Psychological impairment,
described as severe or very severe, was found in almost 40% of the studies. The observed severity
of impairment was most saliently associated with the extent of exposure. In addition, females,
middle-aged persons, ethnic minorities, and the poor were at particular risk for psychological
impairment.
Galea et al. (2005) reviewed studies conducted between 1980 and 2003 and confirm that the
prevalence of PTSD among disaster victims is mostly dependent on exposure to the trauma; that is,
the highest levels of PTSD were found among direct victims of the disaster, whereas the lowest
levels were found among the general population. Although the catastrophic effects of terror attacks
parallel those caused by other disasters, terror attacks differ from natural and technological
disasters in such aspects as their randomness and their intended harm (Dougall et al., 2005).
Finally, in a meta-analytic review of 68 studies encompassing a range of personal traumas (Ozer,
Best, Lipsey, & Weiss, 2003), perceived life threat was found to be positively related to PTSD. A
more recent review of the literature by Ben-Zur and Zeidner (2009) looked at correlational and
experimental studies published between 1999 and 2009. The review was designed to shed light on
the nexus of associations between threat to life and risky behaviors.
Given the massive body of literature on the topic at hand, our coverage of the material is
limited to the concepts that are important for attitudinal forecasting, in particular risk perception
and ABC.
8
Risk Perception and Risk Taking Behavior
Risk is inherent in most types of disasters. A number of risk dimensions have been identified
in past research, including risk perceptions, risk assessments, risky decisions, and risk-taking
behaviors. All risk concepts have one element in common; a distinction between reality and
possibility (Sjöberg, Moen, & Rundmo, 2004). Rosa (2003: 56) defined risk as “a situation or an
event where something of human value (including humans themselves) is at stake and where the
outcome is uncertain.” Risk assessment is based on thoughts, beliefs and constructs (Sjöberg, 1979;
Plapp & Werner, 2006). It is usually identified through experience and available information,
without the need to refer to data and models (Västfjäll, Peters, & Slovic, 2008). The underlying
processes in perceiving risk pass unconsciously. Because risks are low probability events, the
subjective probabilities are usually overestimated (Wickens & Hollands, 2000). This makes it
difficult to formulate a viable theory of risk perception.
In the present study, we consider risk perception and risk-taking behavior as crucial in
disaster situations. Risk perception is the subjective assessment of the probability of a specified
type of accident happening and how concerned we are with the consequences. To perceive risk
includes evaluations of the probability as well as the consequences of a negative outcome.
Perception of risk goes beyond the individual; it is also a social and cultural construct reflecting
values, symbols, history, and ideology (Weinstein, 1989).
Risk perception is affected by familiarity with the source of danger, control over the situation,
and the dramatic character of the events (Plapp & Werner, 2006). Risk-taking behavior has been
defined as “action (or inaction) that entails a chance of loss” (Furby & Beyth-Marom, 1992, p.3).
If disaster situations are indeed one of the antecedents of risk taking, then the degree of risk taking
should vary as a function of the level of exposure to disaster situations, the perceived severity of
the event, the personal consequences of the disaster situation, or some combination thereof.
People experience risk in three ways: (a) risk as analysis, (b) risk as feelings, and (c) risk as
politics (Slovic et al., 2002). Risk as analysis requires the use of logic, reasoning, and scientific
deliberation. Risk as affect brings about fast, instinctive, and intuitive reactions to danger in hazard
management. These approaches of perceiving risk can be used to formulate an experiential systems
analysis and an analytic systems analysis (Slovic et al., 2004). The experiential system is affective
and intuitive. It is fast, mostly automatic, and not very accessible to conscious awareness. The
analytic system supports analysis using algorithms and rules, including formal logic, and risk
assessment. Although analysis is important in some decision-making process, dependence on
9
affect and emotion is quicker and easier. This is because affective reactions are immediate and
automatic, and guide information processing and judgment (Loewenstein et al., 2001). Such
reactions are the basis for decision making heuristics.
Loewenstein et al. (2001) categorized affective responses into anticipatory emotions and
anticipated outcomes, including emotions. Anticipatory emotions are immediate visceral reactions
such as fear, anxiety and dread. Anticipated emotions are expected to be experienced in the future
(Loewenstein et al., 2001), see Figure 2.1.
Figure 2.1 Risk-as-Feelings Perspective (adapted after Loewenstein et al., 2001)
Above all, risk perception is a construct embedded into and determined by society and culture.
According to Douglas (1978), risk perception is not governed by personality traits, needs,
preferences, and properties of the risk objects. It is a culturally, constructed phenomenon (Oltedal
et al., 2004). What is perceived as dangerous and how much risk to accept is a function of one’s
culture and social learning. There are several patterns of cultural differences in communities,
including: (a) communication styles, (b) attitudes toward incidents, (c) approaches to task
performance, (d) decision making styles, (e) attitudes toward disclosure, and (f) approaches to
learning, knowing, and knowledge (DuPraw & Axner, 1997; Banbury & Tremblay, 2004; Klein et
al., 2006).
Affect, Behavior and Cognition
In Chapter 1 we provided the tripartite model of attitude (see Figure 1.1). This trichotomy of
affect (feeling), behavior (acting) and cognition (knowing) can be traced back to Greek
philosophers and was considered in some of the earliest social psychology writings (Breckler,
1984).
Anticipated outcomes (including emotions)
Subjective
probabilities
Other factors e.g.
vividness, immediacy,
mood
Cognitive
evaluation
Feelings
Outcomes:
Behavior
Emotions
10
Affect refers to human emotions or instinct such as anger, happiness, sadness; it also
represents sensory experiences (physical feelings). As a prime determinant of subjective well-
being, affect plays a central role in many activities. Generally, people are imprecise about their
feelings as they are confused between affect and cognition. Therefore, they often rely on feelings
for doing or not doing something (Berger, 2002).
Behaviors are overt, observable responses and actions. They are measurable, and therefore
more easily identified than affect or cognition (Berger, 2002). Consequently, an appropriate
behavior needs to be in place, altered, or diminished in order to reach a particular goal, especially
in managing risk. Some behaviors are much more difficult to develop or change than others,
especially when the behaviors are derived from cultural values (Oltedal et al., 2004). Once
identified, it becomes clearer what needs to be done and what is within a person’s reach.
Cognition encompasses human beliefs, values, decision-making, and perceptions of self,
others, and the world. These perceptions and beliefs include efficacy beliefs, perceptions of locus
of control, and expectations.
Affect and Cognition
In situations such as a disaster, people perceive reality in at least two ways; one is affective
(intuitive and experiential) and the other is cognitive (analytical and rational) (Epstein, 1994).
Formal decision making relies on the analytical and cognitive abilities; unfortunately, this mode is
slow. The experiential/affective system is fast. When a person responds to an event, there is an
automatic search and event matching with the experiential system. This is like searching a memory
bank for related events, including the emotional contents (Epstein, 1994). Figure 2.2 shows the
coupling of affect and cognition.
11
Figure 2.2 Coupling of affect and cognition
Affect is used to evaluate and judge, while cognition is used to interpret and make sense of
the objects, and understand the user (Norman, Ortony, & Russell, 2003). Emotions do not cause
thinking to be non-rational; they can motivate a passionate concern for objectivity. Rational
thinking entails feelings, and affective thinking entails cognition. Rational thinking is more precise,
comprehensive, and insightful than non-rational thinking. However, it is also emotional.
New breakthroughs in neuroscience using functional Magnetic Resonance Imaging (fMRI)
validate the assertions that cognition and emotions are unified and contribute to the control of
thought and behaviour conjointly and equally (LeDoux, 2003; 1995). Additionally, cognition
contributes to the regulation of emotion. Contemporary views in artificial intelligence are also
embracing an integrated view of emotion and cognition. There is evidence to suggest that affect
and cognition interact in many ways, as revealed in behavioral, psychophysiological and neural
measures (Ben-Zur & Zeidner, 2009; Västfjäll1, Peters & Slovic, 2008; Clore & Huntsinger, 2007;
Lerner, & Gonzalez, 2005).
Studies of formal risk analysis have viewed affective responses to risk as irrational (Slovic et
al., 2004). However, analytic reasoning too cannot be rational unless it is guided by emotion and
affect. This is because the analytic and the experiential systems operate in parallel and each
depends on the other for guidance (Levine, 2009). Separate circuits and structures are responsible
for more emotional versus more cognitive functions (LeDoux, 2000).
12
Several studies also examined the relationships between positive and negative affect and
cognitions in those with anxiety and mood disorders (Yamokoski, 2006). The cognitive correlates
of negative and positive affect were studied, and it was hypothesized that worry was related to
negative affect since worry is a cognitive component of both anxiety and depressive disorders
(Beck et al., 2001).
There is also evidence to suggest that emotions, far from being an obstacle to appropriate
decision-making (as they have been viewed from the rationalist perspective), are indispensable for
decision making (LeDoux, 1995). Current neurobiological research is confirming that the heart has
reasons that reason ignores. This does not mean that the emotions cannot make mistakes. Nor does
it mean that certain very strong emotions (specifically those more primary ones which, due to the
nature of the stimulus) may not give rise to unpremeditated acts with disastrous implications for
the person’s life. It is precisely in such cases that the decision process does not take place, as it is
swept aside by more primitive mechanisms (direct routes from the thalamus to the amygdala) that
hijacks the capacity for decision (Damasio, 1995; Davis, 1992; Simón, 1998). Table 2.1
distinguishes between affect and cognition, and their influences in stimuli evaluation.
Table 2.1. Distinctions between Affect and Cognition
Affect and Emotions Cognition
Judge the preferences Evaluate the correctness
Instigation to action Less linked to behaviour
Expression are fairly universal across cultures Restrict to certain language, not cross-culture
Reactions are primary and could happen without cognition awareness
Required sufficient and subjective information input, and accompanied by affect
Affective evaluation of stimuli involves a low level of arousal
Cognitive interpretation involves a deeper understanding of stimuli and human behaviour
The new frontier of knowledge about emotions is modifying our view of the relationship
between the cognitive and affective worlds of the human being. The integrated view is further
explored in the concept of citarasa (Khalid et al., 2010a). Originated from Sanskrit, the term is
widely used in the Malay culture and worldview to explicitly express emotional intent. The word
“cita” means ‘hope, desire,’ and “rasa” means ‘taste, feelings.’ Citarasa, in the context of human
experience, synergistically integrates emotions and cognition in manifesting behavior intent. While
13
there is insufficient evidence in the literature to support citarasa as a concept, there is ample
evidence from fMRI studies of the emotion-cognition connectivity (LeDoux, 2003; Dolan, 2002).
Affect and Behavior
Fear causes fleeing and thereby saves lives: this exemplifies a popular and common sense
view. However, it is increasingly untenable that the primary function of emotion is the direct
causation of behavior. Instead Baumeister et al. (2007) proposed a theory of emotion as a feedback
system which has an indirect influence on behavior. The word emotion suggests a conceptual link
between affect and action control (behavior): etymologically, emotion is from the Latin ex (away,
out) and movere (to move) (Reis & Gray, 2009). Emotions provide feedback and stimulating
retrospective appraisal of actions, and thereby promote learning and change behavior.
Behavior may also be chosen to pursue (or avoid) anticipated emotional outcomes. Rapid,
automatic affective responses (in contrast to the full-blown conscious emotions) inform cognition
and help to guide behavior. Automatic affective responses may also remind a person of past
emotional outcomes and will help in predicting the emotional outcomes of an action.
Empirically, a link between at least some emotions and some actions is unequivocal—in
facial expressions of joy, fear, sadness, for example. Emotion theorists have long posited a major
role for affect in the control of behavior more generally (e.g., Frijda, 1986; Lang, Bradley, &
Cuthbert, 1990; Carver, Sutton, & Scheier, 2000; Schneirla, 1959). Action-control theorists have
equally noted the relevance of affect for action control, including motivation and volition
(Gollwitzer, 1999).
Cognition and Behavior
The theory of reasoned action (Ajzen, 1985) assumes that individuals consider a behavior’s
consequences before performing the particular behavior. As a result, intention is an important
factor in determining behavior. Intentions develop from an individual’s perception of a behavior as
positive or negative together with the individual’s impression of the way their society perceives the
same behavior. Thus, personal attitude and social pressure shape intention, which is essential to
performance of a behavior (Ajzen 1985).
Recognition-Primed Decision model (Klein, 1997) suggests that decisions in naturalistic
situations, characterized by high uncertainty, time pressure and high stakes, trigger appropriate
action schema. If recognition fails initially, humans resort to greater situation assessment until
recognition can occur, eventually constructing a new sequence of actions if an appropriate schema
is not available.
14
Attitudinal Forecasting Model
Affective, cognitive and behavioral forecasting models exist in the literature as standalone
models (e.g. Dunn et al. 2007; Hoerger et al., 2009). Attitude forecasting model that encompasses
all three components has yet to be developed since Breckler (1984). Solomon (1996)
conceptualized ABC components as simultaneous occurrence, as illustrated in Figure 2.3. His
model is built upon Albert Ellis’ model (see Wirga & Bernardi, 2008), to explain the relations
between the environment situation with emotional and behavioral reactions.
Figure 2.3 ABC Model of Attitude (Solomon, 1996). The three components are simultaneous.
The challenge is measuring these components that impact on attitude. Various measurement
tools have been constructed to obtain attitude information. Measures may include the use of
physiological cues like facial expressions, vocal changes, and other body rate measures (Breckler
& Wiggins, 1992). For instance, fear is associated with raised eyebrows, increased heart rate and
increase body tension (Dillard, 1994). Other methods include concept or network mapping and
using primes or word cues (Shavelson & Stanton, 1975).
Human Factors in Disaster Management
The literature on disaster management (DM) identifies several factors that are involved in
disaster response and recovery, but we classify them into four, namely: People, Equipment, Task,
and Environment (PETE) factors:
People may be individuals or groups that are involved in the disaster either directly as
survivors, or indirectly as members of a service team. The latter may be involved in changing
the evolving situation, but at the same time they are exposed to the consequences of the
situation. People may be trained to apply existing plans (e.g. medical personnel, policemen,
soldiers, firemen), but some are not (e.g. volunteers, citizens, journalists), and their actions are
usually reactive rather than anticipatory. Survivors may be proactive if trained or reactive if
not.
Affect
Behaviour
Cognition
Feeling
Intention
Beliefs
Attitudes
15
Equipments include tools and methods for managing the crisis (e.g. communication networks,
transportation means and infrastructures, tools and spare parts for repair and first aid). These
resources are limited and may be seriously damaged in the disaster. Objects used by survivors
to save their lives may be classified as endurance tools.
Tasks are actions undertaken by people using available technologies (e.g. inform public,
remove public, provide first aid, remove injured people, stop spread of the danger or its
consequences, repair infrastructures). These tasks may be interdependent and coordination
between people is required. Furthermore, some pre-planned tasks may be difficult to carry out
due to damaged equipment and limited resources; equipment may also need to be modified for
the purpose. Actions of survivors may be categorized through their Affect, Behavior and
Cognition (ABC).
Environments include physical and social conditions. Physical environment refers to natural
elements such as water, earth, fire, while social environment refers to organisations implicated
in the disaster management system (e.g. police, fire brigade, army, government, hospitals).
They may be public or private. Survivors, for example, interface with both natural and social
environments for life-saving support.
In the present research, the focus was People and Environment, as illustrated in Figure 2.4.
The model is also called Citarasa Behavioral Model (Khalid et al., 2008). There are two sub-
systems: Environment and People.
Figure 2.4 People-Environment Systems model
16
Environment. This subsystem documents information derived from sources that are
associated with hazardous risks, namely, Disaster Situations, Service Organizations and Risk
Culture. The Disaster Situation comprises various disaster types (e.g., natural disaster –
tsunami, earthquake, etc, human induced disaster – terrorist attack, fire, etc., pandemic – swine
flu, HIV, etc) that trigger the organization of Services (e.g., emergency, rescue) efforts that can
be driven by experts (e.g. medical doctors), teams (e.g., military aid), and/or community. The
Cultural values of people in terms of their risk assessment and situation awareness play
important roles in influencing their risk attitudes. However, these measures are not available
in the present study.
People. This subsystem describes the interaction between affect and cognitive systems,
resulting in behavior. The cognitive system follows a human information processing model.
Perception of risks triggers the belief system that includes cognition and memory recall, which
involves appraisal and action (Sjöberg, 1996). The affective system generates a variety of
emotional feelings such as fear, despair, anxiety and disorientation, depending on the threat to
life (Ben-Zur & Zeidner, 2009). The resulting emotional experience predisposes the person to
act. There may be also positive emotional reactions including generosity towards others and
cooperation to help in teamwork (Ebata, 1995). The behavior system forms intention and
outputs overt action that may avert the risk.
Analytical Tools. To analyze the relationship between the Environment and People
subsystems, several methods may be used. Many studies have used psychometrics to analyze
risk perception (Johnson & Tversky, 1984; Lichtenstein, et al., 1978; Slovic, 1987; Slovic, et
al., 1985). People expressed their understanding of diverse hazards through quantitative
judgments of current and desired riskiness (Slovic, 1987). Using multivariate analysis,
relationships are revealed to produce “cognitive maps” or risk attitudes. Shiloh, Guvenc, and
Onkal (2007) studied cognitive and emotional aspects of terror attacks using semantic
differential scale to measure risk perception on 10 negative affect: fear, helplessness,
hopelessness, anger, intolerance, pain, loneliness, insecurity, sadness, and anxiety.
In the context of the present study, text mining was used to identify semantics that
constituted ABC of risk attitudes. In the next study, a Web tool and projective technique will
be used to measure risk attitudes on dimensions such as group culture (e.g. values, norms), risk
assessment, and situation awareness, to be gathered in online and field studies.
ABC Conceptual Model
The conceptual ABC model guiding this research may be described as follows:
1) Disaster situations may influence attitudes of people, including those directly affected and
those not directly affected by the disaster. The first component for attitude forecasting is to
17
identify the specific affect, behavior, cognition (ABC) concepts associated with specific
disasters. This will be based on text mining that result in semantic ontologies.
2) Their affect (feelings) and cognition (thoughts) may impact behavior (actions) they might take.
The effects may be evaluated through experimental manipulations of diagnostic information
such as group cultural values, situation awareness and risk perception using projective
technique and why-why-why probe. This aspect of the model will be explored in future to
enhance the forecasting capability of the attitudinal model.
Research on affective forecasting found that forecasters tend to adopt a relatively analytic
approach in imagining their emotional responses to future events. They extracted the focal event
from the noise of its background and break up the event into its most important parts, devoting
careful attention to those parts that distinguish it from similar events. Problems arise because the
emotional experiences that forecasters are trying to predict may stem from a more holistic response
to events. The impact bias effect, i.e., overestimating the intensity and duration of their own
emotional responses to events is a common problem (Gilbert, Gill, & Wilson, 2002; Hoerger et al.,
2009).
For the purpose of the present study, Figure 2.5 illustrates a conceptual framework for
understanding ABC.
Figure 2.5 Conceptual ABC framework
Affect (Rasa)
Threat to Life
EmotionalExperience
Feelings
Risk Perception Appraisal
BeliefCognition (Cita)
Predisposition
Decisionto Act
Intention
Behavior
(Action)
TIME
FEEDBACK
18
In Figure 2.5, the occurrence of affect, behavior and cognition is contingent upon the type of
disaster and time perspective. For example, with natural disasters such as earthquake there is time
to perceive the risk of behavior which may lead to a delay in action. Therefore, affect (e.g., anxiety)
and cognition (e.g., thoughts) occur before behavior. But with human induced disasters, the
suddenness of the event (e.g. terrorist attack) may induce coping behavior with mixed emotions
such as fear.
The above model is the basis for mapping the relationship of ABC in ontology development.
It is assumed that the relationship varies with the type of disaster experience, with human induced
disasters such as terrorist attack inducing people to act due to trauma, while natural disasters may
induce people to reflect and delay reactions. In the next chapter, the characteristics of disaster
types are described to understand their impact on people.
19
3
Disaster Taxonomy
“Disasters are non-routine events in societies or their larger subsystems (e.g.
regions and communities) that involve conjunctions of physical conditions with
social definitions of human harm and social disruption.”
National Research Council, 2001, p. 3718.
There are many definitions of a disaster (e.g. Turner and Pedgeon, 1997, Denis, 1995, Keller
and Al-Madhari, 1996). The definition used seems to be tied up to the discipline using the term.
Turner and Pedgeon (1997) pointed out that no definition of “disaster” is universally accepted. The
Asian Disaster Reduction Center (2003) defines disaster as: A serious disruption of the functioning
of society, causing widespread human, material or environmental losses which exceed the ability
of affected society to cope using only its own resources. Parker (1992) suggested that the preferred
definition of disaster is: …an unusual natural or man-made event, including an event caused by
failure of technological systems, which temporarily overwhelms the response capacity of human
communities, groups of individuals or natural environments and which causes massive damage,
economic loss, disruption, injury, and/or loss of life. This definition encompasses medical-related
disasters such as tuberculosis and HIV/AIDS cases.
DISASTER CLASSIFICATION
Disasters may result from natural causes (floods, earthquake, hurricane, climatic changes,
etc.), human-made acts (nuclear reactor meltdowns, terrorist attacks, fires, etc.) as well as
pathogenic virus (swine flu, SARS, HIV, tuberculosis). These disasters often have catastrophic
consequences, as evidenced by the grave, immediate, and long-term effects of the tsunami that
impacted the coasts bordering the Indian Ocean in 2005 or the terror attacks on the Twin Towers
on September 11, 2001 (hereafter, 9/11). Such high magnitude disasters often lead to grave loss of
life among members of the community, severe damage of private and community property, and
direct physical injury to survivors and their families, friends, and neighbors (Ben-Hur & Zeidner,
2009).
20
Disasters may be classified as shown in Figure 3.1.
Figure 3.1 Classification of disasters (Shaluf, 2007)
The World Health Organization (2010) and World Health Organization/EHA (1998)
classified disasters as follows:
(1) Natural (physical) disasters: Weather-related disasters (meteorological): – hurricanes, cyclones, tornados, typhoons; – heavy rains, thunderstorms, floods, snowstorms; – drought and famine; and – heat waves, cold waves.
External (topographical): – landslides; and – avalanches.
21
Internal (tectonic and telluric): – earthquakes; and – volcanic eruptions.
Natural (biological): – infestations; and – epidemics.
(2) Man-made/technological disasters: – industrial disasters; – nuclear accidents; – chemical accidents; – fires; – wars, civil strife; and – structural failures (dams, mines).
To ensure adequate coverage of disaster types, in this research we classify disasters into 3
main classes: Natural Disasters, Human Induced Disasters and Pandemic Disasters.
Natural Disasters
A natural disaster (ND) is a consequence when a natural hazard (e.g., volcanic eruption or
earthquake) affects humans and/or the built environment. Human vulnerability, caused by the lack
of appropriate emergency management, leads to financial, environmental, or human impact. The
resulting loss depends on the capacity of the population to support or resist the disaster, that is,
their resilience. Therefore, a natural hazard will not result in a natural disaster in areas without
vulnerability, e.g., strong earthquakes in uninhabited areas.
Earthquakes
An earthquake is a sudden break within the upper layers of the earth, sometimes breaking the
surface, resulting in the vibration of the ground, which strong enough will cause the collapse of
buildings and destruction of life and property (Centre for Research on the Epidemiology of
Disasters, 2003). Earthquakes sometimes trigger landslides, avalanches, flash floods, fires, and
huge, destructive ocean waves (tsunamis) (Federal Emergency Management Agency, 2003a, b).
Powerful aftershocks frequently occur, causing further damage and increasing psychological stress
(International Federation of Red Cross and Red Crescent Societies, 2003).
Earthquakes strike with no early warning and can be devastating. Quakes are measured
according to the Richter scale – the most devastating effects are seen at Level 6 and above and if
the epicenter of the earthquake is located in highly populated areas. Earthquakes can cause high
mortality from trauma, asphyxiation, dust inhalation (acute respiratory distress), or exposure to the
environment (i.e. hypothermia) as well as serious destruction of buildings and infrastructure
22
(International Federation of Red Cross and Red Crescent Societies, 2003; Songer, 200; Schneid &
Collins, 2001).
Tsunami
Tsunamis are a series of large waves generated by sudden displacement of seawater by
earthquakes or volcanic eruption, capable of propagation over large distances and causing a
destructive surge on reaching land (Centre for Research on the Epidemiology of Disasters, 2003).
Tsunamis can originate hundreds or even thousands of miles away from coastal areas. Local
geography may intensify the effect of a tsunami. The areas at greatest risk are less than 50 feet
above sea level and within one mile of the shoreline. Tsunamis reaching heights of more than 100
feet have been recorded. As the waves approach the shallow coastal waters, they appear normal
and their speed decreases. Then as the tsunami hit coastlines, it may grow to a great height and
smash into the shore, causing much destruction.
On 26 December 2004, an earthquake measuring 9.0 on the Richter scale struck the area off
the Western coast of Northern Sumatra, triggering massive tidal waves, or tsunamis, that inundated
coastal areas in countries all around the Indian Ocean rim, from Indonesia to Somalia. At least
150,000 people died in the disaster, with over 525,000 injured, 1,600,000 displaced and over
1,000,000 homeless (International Federation of Red Cross and Red Crescent Societies, 2005).
Hurricane
Hurricanes are large-scale closed circulation systems in the atmosphere above the Indian
Ocean and South Pacific with low parametric pressure and strong winds that rotate clockwise.
There is a maximum wind speed of 64 knots or more for the Western Atlantic and Eastern Pacific
(Centre for Research on the Epidemiology of Disasters, 2003). The primary health hazard from
hurricanes lies in the risk of drowning from the storm surge associated with the landfall of the
storm. Most deaths associated with hurricanes are drowning deaths. Secondarily, a hazard exists
for injuries from flying debris due to high winds.
Hurricanes can be predicted several days in advance. Their onset is extensive and often very
destructive. These disasters are usually more destructive than floods (International Federation of
Red Cross and Red Crescent Societies, 2003). Hurricanes are expected within 24 hours or less. In
regions where hurricanes may occur (e.g. Miami, Florida, in the USA), the National Weather
Service issues information concerning the hurricane. The National Hurricane Center monitors
hurricanes and broadcasts the information needed to track storms. The primary prevention strategy
in the event of a hurricane is to provide early warning and evacuation.
23
Floods
A flood is a significant rise in the water level in a stream, lake, reservoir or coastal region
(Centre for Research on the Epidemiology of Disasters, 2010). Floods can be predicted in advance,
except in the case of flash floods. Flash floods are a sudden and extreme volume of water that
flows rapidly and causes inundation. Because of their rapid nature flash floods are difficult to
forecast and give people little time to escape or to take food and other essentials with them
(International Federation of Red Cross and Red Crescent Societies, 2010).
The impact of flooding can include the destruction of housing, crops, cattle and people. The
primary hazard from flooding is drowning. This is particularly evident for flash floods. A longer-
term health concern from flooding is the development of disease from inundated sanitation stations.
Large floods pose a hazard to existing sanitation and drinking water systems (Shaluf, 2007).
Factors influencing the severity of the hazard are the depth of water, the duration and velocity of
the flood, the rate at which the water rises, and the frequency of occurrence and season (World
Health Organization, 2010).
Human Induced Disasters
A human induced (HID) or human-made disaster has an element of human intent, negligence,
or error that causes the disaster or involving a failure of a human-machine system.
Terrorist Attacks
Terrorist attacks or terrorism acts describe a variety of violent acts which are intended to
create fear, to carry out holy duty, political or ideological goal, and to get immediate publicity for
their causes. Acts of terrorism can range from threats to kidnapping, airline hijacking, suicide
bombers, explosions, mailings of dangerous substances, assassination, cyber attacks and the use of
weapons of mass destruction such as chemical and nuclear. In various definitions of terrorism, the
common thread is politically motivated, often directed at non-combatants and committed by sub-
national groups or underground agents (Ruby, 2002).
Militant groups seeking to overthrow authority have frequently used exemplary violence such
as hostage-taking and bomb scares to intimidate political opponents. These violent acts are also
often directed at non-combatants, that is, people who are not members of the military services or
military members who are not actively involved in military hostilities (Borum, 2004). This
criterion identifies terrorism as violence directed toward civilian populations or groups who are not
prepared to defend against political violence. This also includes military members who are
attacked during peacetime (Ruby, 2002).
24
In addition to the political motivation of the acts, terrorism is intended to create an extremely
fearful state of mind. This fearful state is not intended for the terrorist victims but it is actually
intended for a group or individual who may not have any relationship with the victims (Kaplan,
1981). Therefore, terrorist attacks can be without warning to the general population.
Fire
Although a fire disaster need not necessarily reach catastrophic proportions, it will present
some of the characteristic aspects of a disaster because of the highly destructive action of fire and
of the considerable number of victims. The surviving casualties will have mainly serious and
extensive burns requiring immediate rescue procedures that cannot always be provided by local
resources. A fire of vast proportions can moreover cause damage to the surrounding environment
by the massive production of heat and the emanation of burn gases and fumes.
One factor that makes all fire disasters dramatic is panic. Anybody close to a sudden fire is
affected by panic. This is due to the realization that the fire can kill within a few moments, cause
injuries and permanent disfigurement, and inexorably destroy everything in the vicinity. When a
violent fire breaks out, there is an initial moment of psychological paralysis, generally followed by
total incapacity for logical thought, and this leads to instinctive behavioral reactions whose one
aim is to save oneself and reach safety.
Industrial Accidents
Industrial accidents are mishaps in an industrial or manufacturing setting, possibly caused by
improper work practices, negligence or incompetence, which results in personal injury and/or
property damage. The accidents can also originate from technological failures, dangerous
procedures, infrastructure failures, certain human activities which may cause the loss of life or
injury, property damage, social and economic disruption or environmental degradation.
Technological disasters include accidental release of pollutants that pollutes the environment,
chemical explosion, nuclear explosion, mine explosion and oil spills.
Transport Accidents
Transport accidents are incidents involving mechanised modes of transport. It comprises of
four disaster subsets: accidents involving air, boat, rail transport and accidents involving motor
vehicles on roads and tracks. An estimated 30 million people have been killed in road crashes
since the invention of the motor vehicle over a century ago. The World Health Organization
(WHO) forecasts road crashes will be the third most important cause of death and disability
worldwide, by 2020.
25
Pandemic Disasters
A pandemic is an epidemic of infectious disease that is spreading through human populations
across a large region, for instance a continent, or even worldwide.
Swine Flu
Swine influenza (also called pig influenza, swine flu, hog flu and pig flu) is an infection by
any one of several types of swine influenza virus. Swine influenza virus (SIV) or S-OIV (swine-
origin influenza virus) is any strain of the influenza family of viruses that is endemic in pigs. As of
2009, the known SIV strains include influenza C and the subtypes of influenza A known as H1N1,
H1N2, H3N1, H3N2, and H2N3.
Swine influenza virus is common throughout pig populations worldwide. Transmission of
the virus from pigs to humans is not common and does not always lead to human influenza, often
resulting only in the production of antibodies in the blood. Symptoms of swine flu in humans are
similar to those of influenza-like illness in general, namely chills, fever, sore throat, muscle pains,
severe headache, coughing, weakness and general discomfort. In August 2010 the World Health
Organization declared the swine flu pandemic officially over.
SARS
Severe Acute Respiratory Syndrome (SARS) is a respiratory disease in humans which is
caused by the SARS coronavirus (SARS-CoV), characterized by fever and coughing or difficulty
breathing or hypoxia which can be fatal. It is new and highly contagious form of atypical
pneumonia. Between the months of November 2002 and July 2003, there were 8,096 known
infected cases and 774 confirmed human deaths (a case-fatality rate of 9.6%) worldwide. Rapid
action by national and international health authorities such as the World Health Organization
helped to slow transmission and eventually broke the chain of transmission. However, the disease
has not been eradicated; it may still be present in its natural host reservoirs (animal populations)
and may potentially return into the human population in the future. This warrants monitoring and
reporting of suspicious cases of atypical pneumonia.
HIV/AIDS
HIV/AIDS (Human Immunodeficiency Virus/Acquired Immune Deficiency Syndrome) is a
disease of the human immune system caused by the virus. Since the AIDS epidemic began, 25
million people have died and more than 40 million are now living with HIV and AIDS. In 2001
26
alone, five million people became infected worldwide. HIV/AIDS has an unprecedented negative
impact on the social and economic development of nations. While 70 per cent of HIV-infected
people live in sub-Saharan Africa, AIDS is a global problem. HIV infection is also spreading
rapidly in South East Asia.
AIDS can be prevented. Individuals and communities can cope with the spread of HIV/AIDS
by being properly informed, assessing accurately the factors that put them at risk of infection and
by subsequently acting to reduce those risks. The problem, according to the World Bank, is that
there has not been sufficient amount of coordinated activities to slow and eventually reverse the
spread of the disease. Individuals, governments, civil society, private sector groups, international
and non-governmental organizations have to fully commit and participate in scaling up response.
Tuberculosis
Tuberculosis (Tubercles Bacillus, TB) is a common and often deadly infectious disease
caused by mycobacterium, characterized by the expectoration of mucus and sputum, fever, weight
loss, and chest pain. Tuberculosis usually attacks the lungs but can also affect other parts of the
body. Tuberculosis (TB) is the single most deadly infectious disease and kills two million people
each year. Of the eight million new cases annually, 95 per cent are in developing countries. Asia
and sub-Saharan Africa are the hardest hit, but Eastern Europe has recently seen a major increase
in the incidence and deaths related to TB after many years of steady decline.
In 1993, the World Health Organisation (WHO) launched the Stop TB initiative in response
to the growing crisis. Since 1998, the International Federation of Red Cross and Red Crescent
Societies has been working closely with National Societies and the WHO to control the TB
epidemic. These efforts were to provide assistance to patients and their families and have shown
for an urgent need to scale up activities in order to have a greater impact on the epidemic
worldwide.
Consequences of Disasters
Today, researchers have sought to understand the sources of disaster vulnerability, in terms of
the societal processes that create vulnerability, how vulnerability is distributed unequally across
societies, communities and social groups, how vulnerability changes over time, and how and why
these changes come about (Tierney at al., 2001: 22). There are two types of vulnerability: physical
vulnerability represents threats to physical structures and infrastructures, the natural environment,
and related economic losses. Social vulnerability represents threats to the well-being of human
27
populations (e.g. deaths, injuries, other medical impact, disruption of behavior and system
functioning) and related economic losses (Cutter et al., 2003; Buckle, 2004).
A recent United Nations report (2008) suggests that natural disasters have devastating
outcomes in terms of loss of human life. Between 2001 and 2005, 483,322 deaths were reported
worldwide as being caused by natural disasters (e.g., tsunamis, earthquakes). Other serious
outcomes of disasters include the property loss, forced dislocation, and instant poverty of the
survivors.
Over the past years, terror attacks ─ constituting a grave threat to an entire community or
nation ─ have become a salient and pernicious form of communal disaster situations (Raviv, et al.,
2000). Terrorism and related acts of political violence are responsible for the suffering and death
of thousands of people across the globe. The RAND and MIPT analyses identified 19,828 terrorist
acts worldwide for the period beginning 1968 and ending 2004. These acts of political violence
resulted in 86,568 injuries—of which 25,408 were fatal (Bogen & Jones, 2006). Such acts of
violence have also been reported as having indirect, vicarious effects on the populace via the mass
media (Dougall, Hayward, & Baum, 2005).
Research has shown that humans have a natural capacity to recover and even to enhance their
individual adaptive capacities under adverse conditions (Miller, 2005). There is evidence to show
that resilient responding is a common, if not predominant, response pattern to disaster (Bonanno,
2004, 2005). It would be valuable for disaster management to more formally build on this
knowledge. Resilience implies the ability to bounce back and even to grow in the face of threats to
survival. Three core principles of resilience are: control, coherence, and connectedness (Reich,
2006).
People need to believe that they have control in their lives, and those who have higher levels
of personal control beliefs have higher life satisfaction (Rodin et al., 1985), morale (Brown &
Granick, 1983), lower levels of depression (Lachman & Weaver, 1998), and they tend to live
longer (Dalgard & Haheim, 1998).
Disasters destroy the familiar, creating behavioral disruption and cognitive disorganization
both for the individual and the society. A sense of coherence, is perhaps less readily apparent on
the surface but may be even more fundamental to individual resilience. People wanted answers.
This drive for knowledge is a fundamental principle in the motivation of all organisms (Fiske &
Maddi, 1961; Berlyne, 1963).
28
A notable characteristic of the behavior of people in disasters is to band together, to seek out
others, to establish bonds even with strangers. A major part of the stressfulness of disasters is the
breaking of stable bonds with other people. Survivors seek to re-establish social bonds as quickly
as possible and suffer anxiety if they cannot. This is one reason why the loss of a family member,
or even the loss of a family pet, is so harmful.
Scope of research on disaster
In the present study, disaster is defined as a crisis where demands exceed capabilities (Khalid
et al., 2010b). Disasters are non-routine events that occur in societies or their larger subsystems
(e.g. regions, communities). They involve conjunctions of physical conditions that create human
harm and social disruption (Quarantelli, 1998). This means that for disasters to occur in time and
social space, both conditions are individually necessary and collectively sufficient.
In general, disasters are characterized by being:
sudden, the onset usually is unforeseen and happens without warning,
serious, the functioning of a community or a society is disrupted,
overwhelming, the human, material, economic and/or environmental losses exceed the
ability of the affected community or society to cope within its own resources,
swift, the development of events following the onset is so rapid that it exceeds the ability
to respond, and finally
surprising, events often develop in unexpected directions, thereby challenging the
readiness and resources of the responding organizations.
The impact of disasters may be loss of life, injuries, displacement of large groups of people,
disease, disability, food insecurity, damage or loss of infrastructure and basic services, serious
societal and environmental threats, weakened or destroyed public administration and reduced
public safety and security.
Orasanu and Connolly (1993) identified eight factors that are common to disasters: (1) ill-
structured problems, (2) uncertain dynamic environments, (3) shifting, ill-defined, or competing
goals, (4) action/feedback loops, (5) time stress, (6) high stakes, (7) multiple players, (8)
organizational goals and norms. All of these characteristics put demands on the ability to
communicate and inform people so that they can organize themselves and respond to the situation
in an orderly fashion.
Therefore, effective management of disasters requires a common understanding of the situation
among the principal groups of people involved. These are: (1) the formal organizations at the
29
national, regional and international level, comprising of disaster management teams, rescue teams,
and government authorities on the one hand, and (2) the public at the individual or community
levels on the other. Since the two principal groups are likely to respond in different ways, there is a
need to understand them in order to build a resilient community (Hollnagel, Woods & Leveson,
2006). Understanding how this “social capital” can be effectively developed and managed can
result in efficient life saving operations (National Research Council, 2006).
Besides the negative impacts of disaster on lives and properties, lessons learnt from them
have led to development of safety laws and establishment of specialized functional organizations
and bodies. Even more significant have been the changes in attitudes regarding the inevitability of
disasters (Toft & Reynolds, 1994), from fatalistic resignation to intervention at gaining human
control over fate (Aini & Fakhrul-Razi, 2010).
The focus in our study is on risk attitudes of people as they narrate or report their experiences
in the following disaster categories:
Natural disasters - tsunamis, earthquakes, hurricane, and floods;
Human-induced disasters - terrorist attacks, fire, industrial accidents, and transport
accidents;
Pandemics disasters - swine flu, SARS, HIV/AIDS, and TB.
The next chapter presents a semantic framework for understanding the relationship of attitudes
and disaster situations.
30
4
Semantics Framework
This chapter presents the semantics for development of ontologies that enable analysis of
attitudes during disasters. The semantics were mined using text mining tools and these are
described in the sections below.
SEMANTIC MAPS FOR VISUALIZING INFORMATION
Semantic maps are graphic representations of the relationships between key concepts. We
assume that there are multiple relations between a concept and the associated knowledge structures.
For any concept there are at least three types of associations (Estes, 1999):
associations of class − the order of things the concept falls into (e.g., themes);
associations of property – attributes that define the concept (e.g., requirements, classifier);
associations of example − exemplars of the concept (e.g., attitudinal constructs, data type).
Semantic processing involves an integration of information in a semantic map. The semantic
value of the map becomes apparent when it is relevant to a personal experience, so that the
information can be processed in a meaningful way. By using a semantic map, an individual can
map data onto a visual structure (Card, 2003).
In this study we used semantic maps derived from meaningful personal experiences, to
develop ontologies that would lead to attitudinal forecasting for disaster management.
Ontologies
In modeling attitude for disaster management, there is a need to integrate information from
various heterogenous sources. The use of ontology helps to reveal the implicit and hidden
knowledge that are crucial for model development (Wache et al., 2001). Ontology is an explicit
formal specification of a shared conceptualization (Gruber, 1995; Sheth, 1999; Xu & Zlatanova,
2007). Conceptualization is an abstract model of the real world phenomenon, by which the real
world is identified as a set of concepts (Studer et al., 1998); shared means that the notions are
accepted by a certain group as consensual knowledge (Uschold & Gruninger 1996).
31
In defining ontologies, the meanings of the types of concepts that are used in the
conceptualization and the constraints of their usage are made explicit (Studer et al. 1998, Uschold
& Gruninger, 1996). In addition, the ontology is defined formally in an artificial and well-defined
language so that it is machine-readable (Uschold & Gruninger, 1996). As such, ontology could be
defined explicit or implicit, with different degrees of being formal and being shared (Audi, 1995).
In this project, we define ontology as a way of explaining reality by breaking it down into
concepts, relations and rules, to share it with others. In this phase of the project, we will only
specify the ontology architecture for attitudinal modeling using a hybrid approach.
Ontologies approaches
Ontologies may be built from top-down or bottom-up approach. Top-down ontology is
constructed by first examining the domain of interest at an abstract level and then constructing it
based on top levels concepts. Therefore, an abstract model of the domain of interest is built first
and extending the model further to map more specific concepts from low levels. The result is a
generic model of the domain of interest, and it contains the relations between concepts within the
domain of interest.
The bottom-up approach looks at existing text data sources from low levels (themes, seed
concepts, maps, labels, etc.), and develops ontologies for specific disaster and attitude types, and
combining them as a whole. The resulting ontologies from a bottom-up approach focus on the
specifications of the data sources, and capture the relationship between the data sources.
Both approaches have their advantages; we examined the semantics from text corpuses of
narratives and research reports, and we also identified abstract concepts associated with human
behavior in disaster scenarios on the basis of the people-environment systems model (also known
as citarasa behavioral systems model, see Khalid et al., 2008). The resulting ontologies are general
enough to exchange domain information as well as contain enough information for mapping
among different datasets.
Ontologies architecture
Figure 4.1 shows the architecture of attitudinal modeling for disaster management,
comprising generic ontologies of Risk Attitude and Disaster Situations. Within the generic
ontologies are specific ontologies for each ABC type within the risk attitude class and each
32
disaster situation type within a disaster class. The architecture provides a shared vocabulary for
specifying the semantics.
Figure 4.1. Ontology architecture for attitudinal modeling
For the purpose of modeling, the ontology architecture maps the generic attitude ontology and
the generic disaster situational (organizational) ontology.
Attitudinal type ontologies are those that describe Affect, Behavior and Cognition data
sets that are needed for Risk Assessment, to enable prediction on the basis of Situation
Awareness and Cultural Values.
Disaster type ontologies are those that describe the datasets (e.g., tsunami datasets,
terrorism datasets, SARS datasets, and so on) that are needed for disaster situational
forecasting. Besides being useful for the disaster domain, the datasets could also be used
for other domains such as environmental domain, security domain, and so on. With the
specific disaster type ontologies, one data set can be mapped with another (e.g., from
tsunami data to hurricane data, from swine flu to SARS, etc) or several datasets can be
combined (e.g., the combination of tsunami, earthquake, hurricane, flood data) in a
disaster class. In disaster situational forecasting, we have data not only from the existing
33
databases but also from the field (e.g., culture valence, intensity of values, time
perspective, etc.) that need to be processed and combined.
Generic Ontologies
Attitudinal ontology
Figure 4.2 illustrates a generic attitudinal ontology.
Figure 4.2. Generic attitudinal ontology
In the context of disaster, the attitudinal ontology expresses the following concepts:
Attitudinal components of Affect, Behavior and Cognitive describe the emotions, feelings
(affect), intent, action (behavior), perception, thoughts (cognition) through descriptors or
concepts. Examples of affective descriptors are Fear, Trauma, Anger; behavioral
descriptors are Anticipation, Coping, and cognitive descriptors such as Perceive, Sense.
Emotion expresses subjective feelings toward the perceived risk. This may range from
low to high intensity, depending on risk assessment and situation awareness, that may
induce a threat to life. The semantic interpretation of Emotion is based on Affective
34
Descriptors derived from the dataset in conjunction with a level of description or
conformity.
Perception expresses how the risk situation is perceived. This is an entirely subjective
element which can only be interpreted as a conjunction between perceiver and perceived
risk situation. Cognitive Descriptors derived from the datasets enable semantic
interpretation of Perception. The level of description or conformity, such as importance,
severity, determines the requirements.
Action expresses how the risk situation should be acted upon. Action uses Behavioral
descriptors to enable their semantic interpretation. A level of description or conformity can
be defined to give different weight to different Action output.
Disaster situational ontology
Figure 4.3 shows a generic ontology of disaster situation.
Figure 4.3. Generic disaster situational ontology
In Figure 4.3, the disaster situational ontology expresses the following concepts:
Disaster Situation which is described by the descriptors in conjunction with its level of
importance based on the disaster dataset.
35
Situational Needs expresses the requirements in conjunction with the level of conformity
of the dataset. This may range from low to high need, depending on the disaster situation.
Services describe the requirements of help facilities in relation to situational needs and
level of conformity of dataset. The services may be at the level of local, regional and/or
international, depending on the importance of situational requirements.
People describe the actors involved in a disaster situation. They include rescue agencies,
medical team, military, community and the public. Their level of participation is
determined by the nature of the service descriptor and situational needs.
The Requirements express the relationship between the situation and needs, while the Classifiers
define the level of conformity to the requirements.
In sum, the attitudinal ontology describes the attitudes of the people at risk (e.g., victims,
survivors), while the disaster situational ontology describes the organization of services and the
actors (e.g., relief team) involved to support the needs of the people at risk. These ontologies form
the architecture for a disaster management system. To identify the semantics for the ontologies, the
next section describes the process of text mining.
Text Mining
Text mining is a process of extracting meaningful information from a large number of free-
from documents. It uses techniques from data mining, machine learning, natural language
processing (NLP), information retrieval (IR), and knowledge management.
Traditional data mining used highly structured data represented in numbers. However, text is
a highly unstructured data format and difficult to deal with algorithmically. But it does not mean
that the two concepts are distinctive. This is because text can be processed and transformed into a
numerical representation in the mining process, and the same mathematical techniques can be
applied.
Latent Semantic Analysis (LSA)
Latent Semantic Analysis (LSA) is a theory and a method for capturing and analyzing the
similarity of words and text passages by statistical computations applied to a large corpus of text
(Landauer & Dumais, 1997). The baseline theory of LSA is that the aggregate of all the word
contexts in which a given word appears or does not appear provides a set of mutual constraints that
36
largely determines the similarity of meaning of words and sets of words to each other (Landauer,
Foltz & Laham, 1998).
The similarity estimates derived by LSA are not simple co-occurrence counts, but are based
on Singular Value Decomposition (SVD), which is capable of correctly inferring much deeper
relations within text passages. Singular Value Decomposition (SVD) is closely analogous to factor
analysis. LSA believes that the choice of dimensionality in which all of the local word-context
relations are represented can be of great importance and that reducing the dimensionality – the
number of parameters by which a word of passage is described, from the number of initial contexts
to a much smaller number, will often produce much better approximations of human cognitive
relations. It is this dimensionality reduction step, that surface information is captured into a deeper
abstraction and the latent relationship between words and passages is revealed (Landauer, Foltz &
Laham, 1998).
LSA has been applied to a wide range of problems for identifying the contextual meanings of
words and documents, including information retrieval (Dumais, 1994; Foltz, Kintsch, & Landauer,
1998) assessing learning (Landauer, Laham & Foltz, 2000; Landauer & Dumais, 1997), modeling
knowledge acquisition (Landauer & Dumais, 1997), identifying shared understanding in design
(Landauer, Foltz & Laham, 1998, characterizing design team performance (Dumais, 1998) and
predicting team performance on simulated military mission (Foltz, Kintsch, & Landauer, 1998).
A typical latent semantic approach involves five steps: capturing the language-based
communication, transcribing and text processing, creating word-by-document matrix and log-
entropy matrix, applying LSA to log-entropy matrix and constructing the approximate matrix, and
finally computing semantic coherence between team members and visualizing. Figure 4.4
summarizes the flow of the LSA approach.
In the present study, we adapted the typical semantic approach to fit our research purpose.
There were three steps in our approach as outlined in Figure 4.5. After gathering text-based
narratives and written reports (i.e. text corpuses), the text was processed by the Stanford POS
tagger (Toutanova, et al., 2006) to assign parts-of speech to each word. This software used the
Penn Treebank corpus annotation set developed by University of Pennsylvania. This step
generated words that only carried content-bearing information. The meaningful words with parts
of speech information were used to construct the word-by-document matrix. The essential words
were retained, including nouns, verbs, and adjectives. Only words which appeared more than twice
in the utterances were extracted to form the word-by-document matrix, which counted the
frequency of occurrence of each unique word in each utterance.
37
Figure 4.4. Flowchart of a typical latent semantic approach
Figure 4.5 Latent semantic approach of the study
The third step was to represent the text in a matrix format as shown in Figure 4.6. The rows
of the matrix represented all unique words found within the text corpus, and the columns
represented each individual utterance. Each cell contained the frequency of each word as it
appeared in the utterance denoted by its column. The result was expressed in a word-by-document
matrix. From the word-by-document matrix, the number of occurrence of each word was then
calculated.
Capture text‐based narratives and
research communication
Create word of document matrix
and calculate occurrence of words
Assign parts of speech to text
38
2002
0132...
1121
0101
1100
...
3
2
1
21
m
n
Word
Word
Word
Word
DocDocDoc
X
Figure 4.6 Representation of word-by-document matrix
Leximancer Analysis (LXM)
To analyze the text corpus for developing a semantic framework, we used Leximancer, a text
mining tool. Leximancer is a software system for conceptual analysis of text, with the ability to
map text using a set of conceptual dimensions. Leximancer used a classification system of learned
lexical concepts, rather than keywords. It provided automatic analyses (through machine learning)
as well as customized content analyses using pre-defined concept classifiers. This allowed efficient
classification and quantification of large text documents (Smith, 2005).
Themes or concept groupings were identified in the textual data. They represented clusters of
keywords such as phrases and name-like concepts. Using a word coherence technique,
relationships between concepts, within and between different data sources, were identified. The
output was a visual representation of these concepts and relationships, which were used to quantify
and display the conceptual structure of a set of documents.
Although Leximancer can support .pdf and .html files, all the data were converted into a
single .doc format to ensure that all data were analyzed thoroughly. After a set of data was selected,
the data were pre-processed to convert the raw documents into useful format for further processing
by identifying sentences and paragraphs. Once the data had been pre-processed, important
concepts were automatically identified from the text.
The concepts in Leximancer were collections of words that travel together throughout the
text. Therefore, thesaurus learning identified clusters of words that traveled together. Once the
concept identification had been learned, each block of text was located to identify the location of
the concept from the data set. The last phase of processing in Leximancer was mapping, whereby
the conceptual map of the data set was displayed. Figure 4.7 summarized the Leximancer
Analysis process.
39
Figure 4.7. The process of Leximancer analysis
In essence, the text mining was conducted in three stages:
Stage 1: Data Exploration. To understand the content of the data set, Leximancer analyzed
the data freely, producing an initial exploratory map. In this initial analysis, the duplicate text
sensitivity setting was increased to its maximum in order to avoid spurious results, and the number
of concepts function was set at 400, using the default threshold value of 14. At this stage, no
variables were added or removed. The resulting semantic map was then analyzed, and concepts
that were deemed irrelevant to the disaster context were removed from the map.
Stage 2: ABC Classification. The derived concepts from Stage 1 were converted into a
spreadsheet for identification of ABCs by three classifiers with a background in cognitive science
and human factors. Non-ABCs were removed during this stage.
Stage 3: Final analysis. In this stage, concepts that had been identified as ABCs were further
processed by the tool. This time, the Automatic Concept Identification was turned off; instead the
Concept Seed Editing was turned on. This enabled generation of themes and concepts that are
meaningful to the context. In order to generate an overview of the main themes and their groups of
concepts, the theme size was set at 50% and the concept size at 0%. Semantic maps of each
disaster type were compared to identify consistencies in concepts across disasters as well as
variables that differed between contexts.
40
5
Research Methodology
This chapter describes the methodology for data preparation and analysis. The tools that were
used in the study and their processes have been described in Chapter 4. A pilot study to test the
tools was undertaken and reported in this chapter.
TEXT PREPARATION
The procedure for gathering of secondary data and preparation is summarized below.
1. Text corpora were gathered from digital libraries and Web sources. They came from two
sources: personal stories and research articles. The personal stories were obtained from blogs,
while the research articles from e-journals. Appendix A provides a listing of the files that were
used in the study.
2. The material was screened for ABC criterion. This means that only text containing information
about people’s affect, behavior and cognition in relation to a disaster situation was used.
3. The material was converted to txt format for processing by both tools. This is because LSA
could not read files in .pdf format, unlike Leximancer.
4. The specimens were cleaned to remove words that were not relevant to semantic processing.
This included titles of document, names and affiliations of authors, source of publication, and
so forth.
5. Each data file was recorded into a specimen template and the number of words was counted
for each text specimen. Appendix B gives an example of data recording template for a disaster
from each disaster class.
Corpus sample
Table 5.1 summarizes the text corpora used in the study. The total number of words from
narratives (i.e. personal stories) is 308,168 words, while the total from Research Articles is
350,189 words.
41
Table 5.1. Text Corpus by Disaster Class
Disaster Class Disaster Type
Narratives/People’s Stories
Research Articles
No of articles
Corpus count (no. of words)
No of articles
Corpus count (no. of words)
Natural Disasters Tsunami Earthquake Hurricane Flood
27 29 38 31
39,409 77,430 31,742 52,857
13 15 30 14
42,034 59,984 16,521 24,655
Human Induced Disasters
Terrorism Fire Transport Industrial
17 16 9 14
23,936 25,821 10,824 9,618
15 17 11 12
40,205 41,084 23,983 12,311
Pandemic Swine Flu SARS HIV/AIDS Tuberculosis
8 6 19 12
16,108 5,471 7,918 7,034
10 17 18 6
14,656 14,027 32,447 28,912
Total 226 308,168 178 350,819
Example of Disaster Text Corpus
Below we present an example of text for each of the disaster type within a disaster class.
Natural disaster
Tsunami Personal Stories Tsunami Research articles
A cresting wave knocked Sujith down, and he felt pain shoot through one of his hands. He staggered up and, wading through the wash, came upon his father, who had been swept into a ditch and was waist-deep in water. Sujith pulled him out and screamed, "Run! Run!" The words were barely out when another wave carried them both away, hurtling them along like tree stumps caught in rapids. Sujith tried to grab hold of a concrete post that whipped past him, but missed. Instead the wave wrapped him around a palm tree, which he struggled to climb. Reaching the top, he could see that his little community of 11 houses was cut off from the rest of Weligama by a sewage canal that was now transformed into a raging river. He desperately tried to get a glimpse of his family, but all he could see was his house. The water covered the windows and front door. He feared the worst (Dobbs, 2005, p. 4).
Entrusting children to orphanages irrespective of whether they lost only one or both parents has become a source of emotional confusion for some children who cannot come to terms with the idea that poor coping on the part of surviving parents and/or their poor financial situation should be grounds for placing them in orphanages (ICMH, 2006; India Info, 2005). Among the many child psychosocial implications that have emerged in the wake of the tsunami, eating and sleep disorders and fear of the sea have been the most pervasive. Survivor guilt also remains problematic. In the Maldives, for example, children are reported to have become obsessed with feelings of guilt at what they see as their personal failure to hold younger siblings aloft in the water or to keep hold on them when the sea swept back. (Carballo et al., 2006, p. 3)
42
Earthquake Personal Stories Earthquake Research articles
On Monday, December 22, 2003, at 11:15 am, we experienced an earthquake of 6.5 magnitude, for somewhere between 30 -- 45 seconds, which, for an earthquake, is a long time. I was sitting in front of the tv in the den, watching the news and enjoying a bowl of soup. Stephanie was primping in the bathroom, and Carrol was doing something on her computer in her room. I knew instantly what was going on. Then I heard Stephanie and Carrol, and that brought me to their reality. I went to see what was happening with them, and I found Stephanie on the floor, crying and freaked out, with Carrol hunkering over her, protecting her. As it turned out, Carrol knew right away what was going on -- we are both old hands at this, we both have experienced earthquakes before -- but Stephanie has never had such an experience, and she basically went into a panic. She was in the bathroom, disoriented and scared, and was going to run out the back door, but Carrol got to her (Ballew, 2008, p. 1).
The association between fear and earthquake-related psychological distress is consistent with findings from a previous study of survivors of the 1981 earthquake in Athens which compared survivors who had returned in their homes to those who still lived in tents 2–3 months after the earthquake. In that study the survivors who still lived in tents were more anxious and reported (retrospectively) significantly greater fear during the earthquake. The association between fear and post-earthquake psychological distress is also consistent with results from studies based on survivors of the 1999 Marmara earthquake in Turkey (Livanou, 2005, p. 6).
Hurricane Personal Stories Hurricane Research articles
I learned just how rude when a tornado hit my city of Tuscaloosa, Alabama, Dec. 16, 2000, killing 11 people and leaving hundreds homeless. It was my first - so far, my only - occasion covering a disaster, and I was ill-prepared for the emotional states the newspaper's staff went through as the tornado and its aftermath unfolded. This isn't one of the topics you usually cover in journalism school: treating victims with respect and sensitivity, while at the same time getting the information you need (Lee, 2001, p. 1).
Findings extend the previous research on disaster reactions by examining contextual differences and have a number of potentially important implications for social policy. For example, evacuation distance was negatively associated with symptoms and traumatic experiences. Residents of the Mississippi Gulf Coast in this study were less likely than the other groups surveyed to evacuate more than 100 miles away from their residences and were subjected to a greater number of immediate events. These residents also had a greater number of PTSD symptoms. Although the relationship of evacuation distance to symptoms was not uniquely associated when controlling for the number of events experienced, this finding suggests that evacuation, not just to a safer dwelling, but far from the area of projected landfall might be a good public mental health policy (Weems, 2007, p. 3).
Flood Personal Stories Flood Research articles
On the first floor I have a view of the entire village and then really realize the magnitude of what has happened. Everywhere you look is water, water and more water. There are people at the townhall who are not too sure about their relative’s fate. There are also people, almost sure about their relative’s fate but still have hope. There are people who know for sure about their relative’s fate and are not hoping any longer.The Reverend Enkelaar is busy trying to supply everybody with food as far as that may be possible. How it happened I don't know, but there was food, as far as I can remember. Water however was a different story. We had no drinks for 12 hours
Many residents living in drainage areas (such as the Shinkawa River area in Nagoya City) anticipated a flood before the Tokai flood. It is often suggested that Japanese people are very concerned about disasters. However, other studies have found that the level of preparedness of the Japanese for disasters is insufficient. Thus, flood perception and the actual level people take to prepare themselves for floods are not always in agreement. To enhance residents’ preparedness for floods, it is necessary to examine the causes of the disagreement between flood perception and practical measures undertaken by residents (Takao, 2002, p. 5).
43
(Oeveren, 1997, p. 2).
Human Induced Disaster
Terrorism Personal Stories Terrorism Research articles
As we prayed, my first instinct was to pray for the victims — both the living and the dead; to pray for the firefighters and police officers on the scene; to pray for the families of the victims; to pray for President Bush and the members of Congress; to pray for my fiancée who was in New Jersey at a seminar, a mere 15 miles from my daughter and I as a bird would fly, but she may as well have been 1,000 miles away; to pray for my children and ex-wife in the Midwestern U.S.; and to pray for the safety of everyone else in the United States, with an emphasis on those in New York City. I heard someone talking, and since we were at CBS Studios and supposed to be getting interviewed, it shook me back to reality, I guess. Looking quickly at my watch, it was now 9:20 a.m. This was one hell of a way to start a day, I thought (Jackson, 2007, p. 3).
Finally, by virtue of their scale alone, collective traumas become the center of media attention. This not only contributes to the collective nature of the event but also has the effect of putting the field of psychology itself in the spotlight. Psychologists are called on to contribute to the public’s understanding of the trauma through the media, dispersing the apparent word of science to a wide audience. The damage caused by misstatements and faulty conclusions drawn from intuition rather than empirical data can thus multiply beyond an individual client. Such statements may be the only contact many people have with psychology, especially if they experienced the trauma through media coverage. Even brief sound bites could guide the public’s expectations for how they and those around them should be responding, including expressing thoughts and feelings (Seery et al., 2008, p. 4).
Fire Personal Stories Fire Research articles
I have to mention that I have always been a very heavy sleeper. The fire alarm went off, I did not hear it and my Mom thought that it was my Dad's alarm clock before smelling the smoke, so it wasn't good. Outside the house there was a man he was delivering the newspapers to the delivery-boy when he spotted the flames and smoke. He quickly rushed over to the house and found my Mom outside panicking, screaming, "MY BABIES ARE IN THERE, MY BABIES ARE IN THERE!!!" He then ran into the house, (keep in mind that this man has never before been inside that house), directly to the room in which my sister and I were in. He picked both of us up and went to go out the door (Thompson, 2002, p. 2).
In the early stages of a fire, the people in a building typically have to either rely on themselves, or be rescued by others in their immediate vicinity. The assistance of the professional emergency services, for example in the form of rescue operations by firefighters and emergency treatment by paramedics, can only be provided after the first and most important stage of a fire. Human behaviour during this initial phase is, therefore, an important factor in terms of survival. It can be defined as the actions that people take based upon their perception of the situation, their intention to act, and the considerations involved before these actions are carried out. Accordingly, how people behave during an escape is referred to as evacuation behaviour (Kobes et al., 2010, p. 2).
Industrial Accidents Personal Stories Industrial Accidents Research articles
You probably wouldn't let us out of here if you had your way, you'd put up a police cordon, that would calm you down. Stops. Don't try to tell me it's not like that. I lived through it. In those first days… I took my daughter and ran off to Minsk, to my sister. My own sister didn't let us into her home, she had a little baby she was breast-feeding. Can you imagine that? We slept at the train station. I had crazy thoughts. Where should we go? Maybe we should kill ourselves so as not to suffer? That was just in the first days. Everyone started imagining horrible diseases, unimaginable diseases. And I'm a doctor. I
Contrary to the researcher’s expectations, the current findings do not support the notion that mineworkers are more susceptible to fatalistic and defensive attributions than their counterparts in non-mining industries. There was no difference of statistical significance between the two cate-gories of accident victims on their perception of supernat-ural forces and environmental factors as causal factors. Generally, accident causality factors were attributed to workers’ dispositional factors, which implicated workers’ incompetence and/or failure, and thus indicating that both groups of
44
can only guess at what other people were thinking (Block, 2006, p. 2).
victims perceived worker dispositional characteristics as primarily responsible for accident occurrences (Gyekye, 2003, p. 5).
Transport Accidents Personal Stories Transport Accidents Research articles
I unbuckled my seat belt and took my turn in a line that formed spontaneously in the central aisle. I did not even try to pick up my brief case that was lying underneath the front seat because every second mattered. By that time, the entire plane was on fire and I could even feel the heat through the floorboards and soles of my shoes. I was one of the first to reach the exit. But to my dismay, I soon discovered that there was no chute for anyone to slide down. I jumped out without any hesitation, but just like the way the plane came down, mine too was not a smooth landing! I went sprawling on the ground the same way a few other passengers did. Fortunately, the height from which we had to jump was not much (Abeyagunawardene, 2008, p. 2).
"This was the most perfect emergency landing I ever saw in my life," sanitation worker Danita Johnson told a local New York news channel, who had watched it from a nearby pier. Mr Kolodjay, who said that people went silent and started praying when the captain told them brace for impact, hit his head on impact with the water, but said he was "fine." "You've got to give it to the pilot he made a hell of a landing," he added. Some eyewitnesses admitted it could have been much worse, reminiscing about Flight 90 in 1984, when an Air Florida flight crashed into Washington DC's 14 Street Bridge immediately after take-off because of heavy snow, killing all but five of its 79 occupants (Quinn, 2009, p. 2).
Pandemic Disaster
Swine Flu Personal Stories Swine Flu Research articles
My family has a long history of vaccine injuries that are not reported as vaccine injuries. The list includes three cases of GBS in same family. My mom got her H1N1 shot on Monday of last week. She is high risk health care worker..I begged her not to get shot. But she did anyway to protect her residents. She thought she may be getting a cold on Wed. She went into resp and cardiac arrest on Sat. She has been on life support for the last five days. She is not breathing on her own, she is in medically induced heavy sedation/coma. One has something to do with the other..what does it take to get medically advised not to take these damn shots? (Derrico, 2009, p. 1)
The latest reports I have seen seem to indicate that the H1N1 flu is not as virulent as initially feared. Apparently though, the early infections during the 1918 flu epidemic were also not particularly virulent. At this point we don't know how this H1N1 virus will evolve, so it makes sense to keep taking sensible precautions. These facts give us another reason not to overreact: it is possible that infection with the H1N1 virus currently circulating will provide resistance against a possibly more virulent later version (Pennebaker et al., 2009, p. 1).
SARS Personal Stories SARS Research articles
For a week, I was so weak I struggled to make it to the bathroom. I had to hang onto things because I was in a room by myself with the door closed, and if I fell, no one would see me. I couldn't shower; I was just too worried that I might fall. I joke now that my hair had a week-long oil treatment. I knew I wasn't going to die. I'm young, in great health, I've never smoked and I never get sick, so I was optimistic throughout the ordeal. . The only tense moment was when my 15-year-old daughter, Nicole, who was under quarantine at our house with her older sister and my husband, came down with a fever and cough. When I heard about Nicole, I cried. We could talk on the telephone, but I felt so helpless. I couldn't be
For example, one participant said, ‘‘They took me to that quarantine room, only asked me to lie down on the bed, and not to go outside except to the toilet by following the yellow line on the floor. You never knew about the next step and how long you would wait.’’ Struggle with possible SARS diagnosis. Each participant mentioned feeling very afraid of being diagnosed with SARS during the quarantine period and were burdened by this shadow. Finally, all of them were completely relieved when SARS was ruled out. At the same time, they were informed of the quarantine ending. Many participants used ‘‘big stone’’ to describe their laden worry about being diagnosed with SARS (Lan Lin et al., 2004, p. 3).
45
with her. I panicked. She turned out to be fine but it was very frightening (McClelland, 2003, p. 1).
HIV/AIDS Personal Stories HIV/AIDS Research articles
Today, I am celebrating an Anniversary. I don't recall the exact day I found out I was infected with HIV, it was sometime in 1984. The day that stands out most for me is November 28th 1985. On November 28th 1985, I was told by my doctor to inform my family, arrange my finances and funeral, I had six months to live! Soon after, I left my job, friends and family and moved far away so that no one would see me get sick and die...but I didn't die. Twenty-five years later I'm still here. The point is that I missed out on so much during those years when I hid my HIV status and lived far away from family and friends. Years, when I could have been close to those I love but instead I lived in isolation (McIntyre, 2009, p. 2).
The challenges that older persons face as a consequence of being caregivers and losing their adult children are manifold. Financially, older adults struggle as their adult children are no longer able to provide them with support. Furthermore, old age pensions in South Africa are commonly the only source of household income and are unable to cover all living, medical, and schooling expenses for themselves and their dependents. Emotionally and psychologically, older adults suffer from feelings of distress, anxiety, depression, helplessness, and hopelessness as a result of their caregiving responsibilities and loss of loved ones (Boon, 2009, p. 3).
Tuberculosis Personal Stories Tuberculosis Research articles
As we had gathered for a party with colleagues and friends, a phone call came from my doctor. She had a simple message; “you can’t leave the country because it has been noted that you have MDRTB”. I honestly had no idea what that was all about! As I woke in the morning, there was an ambulance outside waiting for me. As the rest of team was on their way to the airport, I was on my way to the hospital, two extremes again. That was marking the beginning of another chapter in my life (Ocaya, 2007, p. 2).
His wish centered on spreading awareness of this deadly form of TB and the images are borne out of Nachtwey's frustrations with the underreporting of what is potentially a global health crisis. His photos tell the grim stories of impending death. In one, a man's suffering is so palpable that it is almost impossible to tear your eyes away from him. Another image shows a woman in a Thai hospital staring vacantly, as if resigned to the fact that death is soon approaching. Yet another shows the look of helplessness on a mother's face faintly reflected in the terrified eyes of her ailing child. And so the images continue, revealing with each click of the mouse a photo that is more haunting than the last (Poltzer, 2008, p. 1).
DATA ANALYSES
The data analyses were performed in 5 stages.
1. Frequency analysis using text mining tools, LSA and Leximancer, was performed on the
prepared data. This generated a listing of words mined by each tool, and analysed with
Analysis of Variance (ANOVA) to determine if the tools differed in semantic processing. A
pilot study was conducted to test the feasibility of the tools on text format.
2. The words were counted in terms of the number of similar and dissimilar words mined.
Cohen’s kappa (1960; 1968) was performed on the data set.
46
3. The words derived from Leximancer analysis only was classified manually into ABC by 3
classifiers with background in cognitive science, and human factors.
4. The classified data were screened by a confederate to resolve conflicting results in the
classification.
5. The inter-classifier reliability, Fleiss Kappa (1971; 1981) test was performed on the ABC
counts.
Pilot study
A pilot study was conducted on a specimen of text from a chapter of the book “Unthinkable”
(Ripley, 2008). The purpose of the pilot was to understand the nature of text preparation for
analysis by both tools. The text was derived from a chapter “FEAR” which contained 11,559
words. The text was scanned and converted to .txt format for analysis by both LSA and LXM. An
extract of the text is shown below.
The Physiology of Fear What does it feel like to face death? What happens in our brains as the ground buckles under our feet? Fear guides our reactions in every stat ion of the survival arc. But we'll consider its effects here, in the beginning of the deliberation phase, because fear is typically at its peak once we've grasped the danger we face. Any deliberation that follows will happen through the prism of fear. People's behavior in a disaster is inexplicable until we understand the effect of fear on the body and mind.
The human fear response looks a lot like the fear response of other animals. So scientists understand fear better than, say, guilt or shame. - Fear is so fundamental," says brain expert Joseph LeDoux. "There are key environmental triggers that will turn it on and well-worked-out responses that help you cope with it. These things have stuck around through zillions of years of evolution."
The first rule of fear is that it is primitive. Consider the fact that our hair stands on end in a terrifying situation. What purpose could that possibly serve? Well, none for us. But scientists believe it may be related to the flashing of feathers in birds or fin extensions in fish, all of which aid in the survival of those creatures. Over the long arc of history, fear has served us very well, and it still does, with some exceptions (Ripley, 2008, p. )
The results of the pilot run are summarized in Table 5.2.
47
Table 5.2 Summary of LSA and LXM functions
Functions LSA LXM Text format Operational on .txt format
only. Can be used to analyze characters, e.g. Mandarin characters.
Operational on .txt, .doc, and .pdf formats. Cannot be used on language characters.
Text count Generated a word count of 849 words from a total of 11,559 words. But the words are a mixture of relevant and irrelevant words.
Generated concept count of 190 concepts from a total of 11,559 words. A threshold can be set for concept generation.
Analytical Assigned parts of speech to word, e.g. Fear/NN (noun), take/VB (verb), etc. This is useful in identifying the semantic type.
Provided a semantic map of the main themes produced by seed concepts.
Relationship Computed correlation of target word to other words in the list.
Computed likelihood (%) of concept, mapped to target word.
From the above, it can be concluded that both tools supported functions that are helpful in
semantic processing, but the text had to be converted to .txt format in order to run on both tools.
This necessitated advance preparation of material in the actual study.
48
6
Study 1. Comparison of Tools in Semantic Mining
This chapter presents Study 1 on determining the ability of Latent Semantic Analysis (LSA) and
Leximancer (LXM) in semantic mining.
RESEARCH DESIGN
Aim and Hypothesis
The objective of this study was to determine if the tools mined the same concepts from the
same corpus. On the basis of the tool’s main task, it was hypothesized that both tools should not
differ in their semantic mining capabilities. However, to develop ontology architecture (see
Chapter 4) requires visualization capability beyond semantic mining.
Method
Text corpus
The data for this analysis was selected to represent each corpus type and disaster class. Table
6.1 shows the word count.
Table 6.1. Word count by text corpus and disaster type
Disaster type Personal Stories Research Articles
Natural: tsunami 39,931 52,694
Human induced: terrorism 16,485 63,317
Pandemic: swine flu 17,096 19,022
Total words 73,512 135,033
49
Although there was a difference between the total counts for each disaster type, the number of
ABCs mined from each disaster class from the combined corpus remained the same as shown in
Table 6.2.
Table 6.2 Total ABC count in Text Corpus by Disaster type
Affect Behavior Cognition Tsunami 82 175 42Terrorism 84 149 46Swine Flu 60 98 25
Data analysis
Univariate ANOVA was performed on the frequencies of ABCs, while Cohen’s Kappa on
inter-classifier reliability and evaluated at the 5% probability level.
Results
Cohen’s Kappa for Inter-classifier reliability
Two classifiers were given three set of derived concepts from three types of disasters namely
tsunami, terrorism and swine flu. Each set contains two type of corpus to be rated which are
research & media articles and personal stories. They were asked to code each concept word as: 0 =
Non ABC, 1 = Affect, 2 = Behavior and 3 = Cognitive. The codes from each classifier are
represented as separate variables in the data set for SPSS analysis.
Tsunami
Table 6.3 provides a summary of classification by two classifiers for tsunami personal stories
(PS).
Table 6.3 Classification of ABC for Tsunami Personal Stories
Count
Classifier 2
Total Affect Behavior Cognition
Classifier 1 Affect 14 9 0 23
Behavior 0 72 0 72
Cognition 0 7 12 19
Total 14 88 12 114
Kappa value .701
The kappa results for PS Tsunami revealed r = 0.701, suggesting substantial agreement. Table 6.4
gives the results of classification for tsunami research articles (RA).
50
Table 6.4 Classification of ABC for Tsunami Research Articles
Count
Classifier 2
Total Affect Behavior Cognition
Classifier 1 Affect 20 2 1 23
Behavior 0 25 1 26
Cognition 1 2 15 18
Total 21 29 17 67
Kappa value .841
The kappa results for RA Tsunami revealed r = 0.841, indicating almost perfect agreement
between them.
Terrorism
Table 6.5 is classification for terrorism PS. The kappa results for PS Terrorism revealed r =
0.722, suggesting substantial agreement.
Table 6.5 Classification of ABC for Terrorism Personal Stories
Count
Classifier 2
Total Affect Behavior Cognition
Classifier 1 Affect 23 13 0 36
Behavior 0 75 0 75
Cognition 0 7 16 23
Total 23 95 16 134
Kappa value .722
Table 6.6 is the summary for Terrorism RA.
Table 6.6 Classification of ABC for Terrorism Research Articles
Count
Classifier 2
Total Affect Behavior Cognition
Classifier 1 Affect 23 2 0 25
Behavior 0 25 0 25
Cognition 1 14 13 28
Total 24 41 13 78
Kappa value .676
The kappa results for RA Terrorism revealed r = 0.676; also indicating substantial agreement
between the classifiers.
51
Swine Flu
Table 6.7 is the results of classification for swine flu PS. The kappa results revealed r = 0.803,
that there is almost perfect agreement.
Table 6.7 Classification of ABC for Swine Flu Personal Stories
Count
Classifier 2
Total Affect Behavior Cognition
Classifier 1 Affect 17 5 1 23
Behavior 0 54 0 54
Cognition 0 3 9 12
Total 17 62 10 89
Kappa value .803
The classification for swine flu RA is summarized in Table 6.8.
Table 6.8 Classification of ABC for Swine Flu Research Articles
Count
Classifier 2
Total Affect Behavior Cognition
Classifier 1 Affect 14 0 0 14
Behavior 0 27 0 27
Cognition 2 2 17 21
Total 16 29 17 62
Kappa value .900
The kappa results for Swine Flu RA revealed r = 0.900, also almost perfect agreement.
ANOVA analysis
The effect of the analysis tools on the frequency of ABCs was determined using ANOVA.
Similar concepts in LSA and LXM were identified and used, where N for Affect = 42, Behavior =
52, Cognition = 14. Table 6.9 summarizes the mean and standard deviation (SD) for each ABC
type.
52
Table 6.9 Mean and SD of ABC mined by LSA and LXM.
TOOL TYPE ABC TYPE Mean Std. Deviation N
LXM Affect 16.14 16.829 42
Behavior 17.56 14.927 52
Cognition 17.00 16.072 14
Total 16.94 15.701 108
LSA Affect 16.45 17.892 42
Behavior 15.85 14.041 52
Cognition 16.50 17.253 14
Total 16.17 15.902 108
Total Affect 16.30 17.264 84
Behavior 16.70 14.446 104
Cognition 16.75 16.363 28
Total 16.55 15.770 216
In Table 6.10, the ANOVA results revealed F (1, 216) = 0.06, n.s. This means that the tools
were equal in semantic mining capability.
Table 6.10 Effect of tool on semantic mining
Source Sum of
Squares df Mean Square F Sig.
TOOL 15.802 1 15.802 .062 .803
ABC 8.871 2 4.435 .017 .983
TOOL * ABC 48.032 2 24.016 .094 .910
Error 53380.644 210 254.194
Total 112639.000 216
Semantic mining of similar and dissimilar words
To illustrate that both tools could mine similar words, Table 6.11 provides a listing of select
words from both analyses, and Table 6.12 dissimilar words.
53
Table 6.11 Similar ABC words mined from a sample of disaster class
Affect Behavior Cognition
LXM LSA LXM LSA LXM LSA
Natural Disaster
PS Tsunami
Feel 30 32 Aid 19 19 Decided 23 24 Hope 8 4 Asked 14 14 Planned 8 8 Lost 31 31 Damage 10 9 Thought 65 67 Loved 10 10 Brought 11 12 Normal 11 12 Experience 9 8 Relief 22 23 Lead 4 3 Safe 10 10 Reports 6 6 Strong 8 8 Take 24 24 Used 9 9 Work 24 16
RA Tsunami
Feel 23 27 Aid 29 38 Decisions 13 13 Hope 12 11 Asked 39 38 Planning 15 13 Lost 80 98 Damage 18 20 Thoughts 14 15 Loved 12 12 Bring 10 10 Normal 11 11 Experience 48 51 Relief 29 37 Lead 7 8 Safe 11 13 Reports 10 14 Strong 17 16 Take 17 19 Use 52 20 Work 45 44
Human Induced Disaster
PS Terrorism
Calm 7 8 Act 5 4 Sense 5 5 Fear 5 4 Attack 20 15 Thoughts 8 7 Slow 3 2 Respond 2 2 Concern 3 3 Changed 4 3 Feeling 5 3 Use 9 9 Lost 4 4 Work 36 28 Personal 4 3
RA Terrorism
Calm 48 28 Act 10 12 Sense 35 38 Fear 73 59 Attack 51 64 Thoughts 19 18Slow 8 7 Respond 48 40 Concern 12 12 Change 36 34 Feelings 34 40 Use 49 22 Lost 7 13 Work 23 29 Personal 26 36
Pandemic Disaster
PS Swine Flu
Care 8 9 Cough 6 5 Mind 4 4 Severe 8 7 Take 34 29 Thought 13 9 Sick 12 15 Sneezing 3 3 Fear 5 5 Confirmed 4 4 Feeling 16 16 Provided 3 3 Feels 3 3 Reports 5 4 Treating 3 3 Work 29 32 Use 4 5 Taken 10 10
RA Swine Flu
Care 18 20 Cough 14 11 Mind 6 5 Severe 12 13 Take 14 11 Thought 10 5 Sick 6 6 Sneezing 26 16 Fear 10 9 Confirmed 7 8 Feelings 10 5 Provide 11 7 Feel 6 6 Report 4 5 Treat 3 3 Work 17 13 Use 11 3 Taken 7 5
54
Table 6.12 Dissimilar ABC words mined
LXM LSA
Affect Behavior Cognition Affect Behavior Cognition
Natural Disaster
PS Tsunami
Felt 27 Started 30 Knew 30 Lucky 23 Manage 25 Recall 12 Looked 29 Waiting 24 Trying 24 Washed 24 Told 23 Wanted 24 Tried 22 Arrived 23 Heard 20 Kept 23 Coming 18 Ran 23 Look 18 Shake 23
RA Tsunami
Emotion 44 Study 40 Mental 80 Feeling 37 Seek 41 Mind 23 Affect 30 Impact 37 Loss 33 Asked 39 Support 37 Needs 31 Wait 35 Risk 21 Trauma 27 Provide 35
Behavior 19
Used 52
Report 8
Human Induced Disaster
PS Terrorism
Glad 16 Goes 36 Belief 16 Feel 11 Asked 15 Knew 14
Walking 31 Kept 15 Decided 12 Work 28 Waiting 13 Told 25 Started 12 Heard 22 Walked 12 Hit 22 Called 11 Looked 17 Tell 11 Taken 17
RA Terrorism
Trauma 53 Study 42 Mental 42 Coping 51 Fight 60 Stress 52 Report 31 Worst 42 Support 57 Distress 46 Called 19 Loss 40 Adjust 55 Afraid 45 Risk 7 Trauma 34 Attack 51 Fear 30 Fearful 21
Pandemic Disaster
PS Swine Flu
Told 27 Closed 18 Concern 8 Woke 9 Felt 21 Receive 17 Scared 6 Given 7 Fine 10 Pain 16 Wait 7 Sore 14 Stay 6 Stay 14 Walked 6 Bad 12 Waited 6 Called 6 Saying 6
RA Swine
Flu
Stress 41 Learned 36 Emotion 11 Grab 16 Perceive 14 Scare 32 Stand 34 Panic 10 Threat 11 Emotion 20 Respond 34 Lay 10 See 28 Dying 10 Studies 23 Asked 10 Seek 21 Contact 10 Report 20 Keep 10
55
Word correlation
On the basis of a target word ‘People’, we examined the correlation between it with other
words. Table 6.13 is the results for LSA of tsunami PS. It can be seen that there were several
behavioral words such as find, screaming, help, waiting, see and running that correlated with
people. There were few cognitive and affective words, such as know and lucky, respectively.
Table 6.13. Correlation of target word ‘people’ with other words in LSA for tsunami personal stories
56
Concept mapping
Table 6.14 shows the percentage of likelihood between target word ‘people’ and concepts in
the same text corpus of tsunami PS. Likewise in LSA, there were more behavioral words such as
injured, died, killed, crying, screaming, bought, and floating. Affective words to describe the
people are chaos and lucky, while there was hardly a cognitive word that mapped to the target
word ‘people’.
Table 6.14 Likelihood of word mapping for theme ‘people’ in Leximancer for Tsunami Personal Stories
57
Semantic maps
The LXM analysis generated semantic maps that may be used to analyze the mapping
relationships of various concepts. Figure 6.1 illustrates a semantic map of the ABC of tsunami.
Figure 6.1 ABC themes and seed concepts of Tsunami Personal Stories
The circles denote the central themes that emerged. The central theme is a collection of
correlated words that are discovered through a learning process of the tool, LXM. The theme is
determined by the strength of the likelihood of the concepts to the theme. The color denotes the
theme ranking, based on the strength of concept counts within the theme. For example, red for
‘thought’ has the highest no. of counts per concept that made up thought, followed by ‘felt’, then
‘hold’. The size of the circle denotes the overall number and density of concepts within the theme.
Each concept is mapped to another to form a map or network of concepts. The mapping
relationship between the concepts is determined by the likelihood value, expressed as percentage
(Leximancer, 2010).
58
In Figure 6.1 there are several behavior concepts within each central theme. The concept
thought (largest collection of words) is linked to mind and wanted (cognition), heard, look, moving,
kept, looked (behavior). Other behavior words include: running, shouting, screaming, crying,
helping, grabbed, pushed, and so forth. These words described the experience of people during a
tsunami that occurred suddenly. Overall, there were few affective words such as pain, calm, lucky
and happy.
Figure 6.2 represents the map for terrorism. There are 7 central themes, with “told’ being the
dominant theme. Likewise tsunami, in terrorist attacks, people tend to act. There are considerable
number of behavior concepts such as told (to stop), stopped, take, moving, work, attack, escape,
walking, running, screamed, looked, called, and so forth. The cognitive concepts were few such as
thought, mind, remember, sense and wanted. There were negative affective concepts such as fear,
shock, lost, sick, horrible, and positive affective experience such as cool, calm, comfort, and hope.
Figure 6.2 ABC themes and seed concepts of Terrorism Personal Stories
59
Figure 6.3 shows the semantics of swine flu.
Figure 6.3 ABC themes and seed concepts of Swine Flu Personal Stories
In Figure 6.3, the information differed greatly from the previous maps of tsunami and
terrorism where the central themes are relatively sporadic. The dominant central theme is ‘work’
followed by ‘tell’. The ABC concepts are quite distributed within the main themes, such as use,
taking/take, sneezing, coughing, started, received (behavior), felt, worry, panic, weary, crazy, hurt
(affect), and thought, decided, suspected, guess, idea, believe (cognition).
In short, the ABC semantics differed with the type of disaster, with significant behavior
concepts in tsunami and terrorist attack compared to swine flu. This suggests that people act in the
face of high risk situations, and their behavior may be governed synergistically by both affect and
cognition.
60
Discussion
The purpose of this study was to examine the capability of LSA and LXM in semantic mining.
From the ANOVA findings, it was confirmed that both tools mined similar words and dissimilar
words. In other words both have similar basic capability and there was no need to use both tools in
the next study.
There were differences between the tools in the following aspects:
First, LSA took a longer time to process (about 2 hours) for the same text corpus, compared to
LXM that took only 20 minutes.
Second, LSA applied parts-of-speech to refine words, but LXM filtered words to identify
meaningful concepts only. This generated a list of concepts rather than mere words.
Third, LSA misclassified some words, while LXM did not. This may be due to the algorithms
used by LSA that could not distinguish some words clearly, especially affective and cognitive
words.
Fourth, LSA generated correlations between target word and other words. LXM indicated the
likelihood of occurrence between target concept and other concepts. For forecasting purposes,
the likelihood value is useful.
Fifth, a major difference is LXM produced semantic maps in the form of themes at the first
level, and seed concepts maps at the node level that can be analyzed further to develop
ontology.
Overall, LXM was easy to use with features that proved useful for developing semantic framework.
The semantic maps, represented by the three examples of tsunami, terrorism and swine flu,
revealed visually the ABCs inherent in these disaster situations.
Conclusion
Both tools mined the same word from the corpuses, but the frequency may differ slightly due
to filtering and calculation algorithms of the tool. On the basis of its added value in terms of
usability, LXM was used in the remaining conceptual mining described in the next chapter 7, and
further validation of ABC concepts.
61
7
Study 2. Attitudinal Semantic Mining
RESEARCH DESIGN
Aim and Hypothesis
The aim of this study was to determine the nature of relationship between attitudinal
components (ABC) and disaster situational factors derived from different corpora. It was assumed
that the attitudinal semantics derived from disaster types correlate significantly with the disaster
class. Since the ABC classified data was a count, Pearson Chi Square analysis was performed on
the combined disaster data within each class, and tested at the 5% significance level. Semantic
processing was performed using Leximancer only in light of its utility, as discussed in Chapter 6.
Method
Text corpus
The sample used in this analysis was the full corpus set as outlined in Table 5.1 (see Chapter
5). The size of the corpus was 308,168 words from narratives and 350,189 words from research
articles, with a total word count of 658,357. This corpus size was considered sufficient for
semantic mining. Personal information and references in the text were deleted.
Procedure
There were five stages of analysis: First, a frequency count was used to derive a listing of
about 400 concepts per disaster type. Second, the concepts were classified manually into ABC by
three classifiers. Third, a confederate checked the outcome to resolve conflicting classification
across classifiers. Fourth, Fleiss Kappa analysis was performed to obtain inter-classifier reliability.
Fifth, Pearson Chi Square was used to test the relationship between ABC and disaster class or
corpus type at 5% significance level.
62
Results and Discussion
Fleiss Kappa inter-classifier reliability
Fleiss Kappa analysis was performed on the manual classifications made by three classifiers.
The purpose was to evaluate the reliability of their ABC classifications prior to using the data in
the semantic analysis.
On average, the results of Fleiss Kappa revealed almost perfect agreement with r = 0.80.
Table 7.1 provides a summary of the results. The individual Fleiss Kappa results for each disaster
type may be viewed from Appendix C.
Table 7.1 Fleiss Kappa Results by Disaster and Corpus type
Disaster type Fleiss Kappa Value, r Agreement Result
Personal Stories
Research Articles
Natural disaster
Tsunami 0.785 0.794 Substantial agreement
Earthquake 0.819 0.835 Almost perfect agreement
Hurricane 0.799 0.751 Substantial agreement
Flood 0.709 0.722 Substantial agreement
Human Induced
Terrorism 0.761 0.695 Substantial agreement
Fire 0.656 0.860 Substantial agreement
Industrial 0.881 0.902 Almost perfect agreement
Transport 0.710 0.835 Substantial agreement
Pandemic
Swine flu 0.869 0.865 Almost perfect agreement
SARS 0.853 0.810 Almost perfect agreement
HIV/Aids 0.898 0.758 Almost perfect agreement
Tuberculosis 0.855 0.773 Almost perfect agreement
ABC association within disaster class
The data in this analysis was based on a combined data set for each disaster class across
corpus type. In other words the four types of disasters were combined within each disaster class.
The hypothesis for test stated that ABCs associated significantly with disaster class. The results
showed mixed associations as summarized in Table 7.2. Therefore, the hypothesis was partially
supported.
63
Table 7.2 ABC Association within Disaster Class
Disaster Class
Pearson Chi Square
Significance level
Cramer’s V Association
Natural disaster
2 = 2.98, df =6 n.s. 0.05 -
Human Induced
2 = 20.72, df =6 p=0.002 0.12 very weak positive
Pandemic 2 = 13.16, df =6 p=0.04 0.10 very weak positive
From Table 7.2, there was highly significant association for human induced disasters and
significant relationship for pandemic, but not for natural disasters. Figures 7.1, 7.2 and 7.3 show
the counts for natural, human induced and pandemic, respectively.
Figure 7.1 ABC count within Natural Disaster class
64
Figure 7.2 ABC count within Human Induced Disaster class
Figure 7.3 ABC count within Pandemic disaster class
The cluster of disasters within Natural disasters (tsunami, earthquake, hurricane, flood) are
quite similar to Human induced disasters (terrorism, fire, vehicle accidents) and pandemic (swine
flu, HIV, TB), see chapter 3 for description of disaster type. For example, the common physical
65
elements that characterize tsunami, hurricane and flood are water and wind. The only difference is
earthquake. These natural disasters are announced with warning signs including tsunami. Other
disaster classes such as human induced may be sudden with little warning. Therefore, the ABCs
have little in common across disaster types, unlike natural disasters.
Across the disaster classes there were more behavior concepts than cognitive or affective. In
tsunami, people tend to act more, than in the other the disaster types. This could be due to the
suddenness of the phenomena, which leave people with little choice and time to think (Carballo,
Heal & Horbaty, 2006). Earthquake, hurricane and flood enable people to prepare (Griffin et al.,
2008). Terrorism and fire demand greater action from the victims than industrial or transport
accidents (Altheide, 2006). With pandemic, the behavior is quite similar except for TB.
Concerning emotion, the feelings associated with natural disasters are almost the same. This
could be due to attribution to “Act of God” (Somasundaram et al. 2003, Grandjean et al., 2008).
Therefore, there is little allocation of blame to others. With human induced disasters, the emotions
are greater for terrorism and vehicle accidents where the source of disaster may be attributed to
humans (Lerner et al., 2003). Human-caused events such as dam collapses and industrial accidents
“represent in the eyes of victims a callousness, carelessness, intentionality, or insensitivity on the
part of others” (Bolin, 1985).There is a greater tendency for people to blame others for causing a
disaster. In this regard the attribution of HIV victims are higher followed by TB. This may be due
to the stigma associated with the disease (Gonzalez et al., 2009).
The cognitive concepts across disaster classes differed greatly for pandemic, especially with
HIV and TB. In the case of latent TB, even without any symptoms, but knowing that an illness lies
dormant in the body has an impact on perceived health (Nyamathi, et al., 2005). Fire also induced
people to think about the traumatic loss of home and loss of extended self after the fire (Lollar,
2010).
ABC association within corpus type
The four types of disaster within a corpus type were combined to generate a single data set.
Thereby tsunami, earthquake, hurricane and flood were combined within the natural disaster class
for PS, and likewise for RA. The hypothesis stated that ABCs correlate significantly with corpus.
The results revealed highly significant associations as summarized in Table 7.3. Therefore, the
hypothesis was supported. Figures 7.4, 7.5 and 7.6 illustrate the ABC distribution for corpus and
disaster class.
66
Table 7.3 ABC Association within Corpus Type
Disaster Class Pearson Chi Square Significance level
Cramer’s V Association
Natural Disaster 2 = 62.14, df=4 p<0.0001 0.33 weak positive
Human Induced 2 = 15.32, df =4 p<0.0001 0.15 very weak positive
Pandemic 2 = 18.71, df =4 p<0.0001 0.18 very weak positive
Figure 7.4 ABC count within Corpus for Natural disaster
From Figure 7.4, it can be seen that PS produced more behavior concepts than RA. PS
describe real experiences of the survivors. The affect and cognition counts are almost the same in
PS. RA produced more affective concepts, perhaps because the research was aimed at
understanding PTSD (Ben-Zur & Zeidner, 2009).
67
Figure 7.5 ABC count within corpus for Human Induced disaster
A much greater number of personal stories was generated for Behaviour then for Affect and
Cognition. There was no difference between the number of affective and cognitive concepts.
Figure 7.6 ABC count within corpus for Pandemic disaster
Figure 7.5 shows more behavior concepts with pandemic PS than RA. It can be concluded
that PS is a good source for understanding human behavior in disaster. Since there was highly
significant associations between corpus and ABCs mined, using both types of corpus is important
in the study of behavior for forecasting attitude.
68
ABC semantics of natural disaster corpus
The ABC data from PS and RA corpora were combined to evaluate the central themes and
underlying concepts. Figure 7.7 shows the central themes that emerged from the analysis.
Figure 7.7 ABC themes of natural disaster class
Four central themes emerged from the ABC concepts, namely: Anticipation, Thought, Started
and Imagine. The dominant theme ‘Anticipation’ suggests that natural disaster situations such as
hurricane and flood come with some pre-warning, and there is opportunity to anticipate the
incident. But in other natural disasters such as tsunami and earthquake, which may be sudden and
surprising, the anticipation may be in the form of assistance provided by service organizations to
the survivors. Figure 7.8 represents the concepts that contributed to the ‘Anticipation’ theme.
69
Figure 7.8 ABC concepts of “anticipation” theme for natural disaster class
In Figure 7.8, the concepts that make up ‘Anticipation’ may be classified as follows: behavior
concepts such as anticipation, concern, and plan; affective concepts such as trauma, fear, and relief;
and cognitive concepts such as memories, perception and decision.
The concept ‘anticipation’ is mapped to concern (behavior), fear (affect), and decision
(cognition). This means that in natural disasters, people do have concern over the situation, which
may induce fear; therefore they need to decide for an action. The fear may increase their
vulnerability that caused them to make decisions.
Another concept ‘risk’ is mapped to perceived (cognition), fears and moods (affect). People
perceive the risk induced by the physical nature of the disasters and it affects their mood as well as
making them fearful of the situation. There is indeed evidence to suggest that people experienced
psychosocial health effects as a result of disasters such as tsunami (Carballo, Heal, & Horbaty,
2006; Raholm, Arman & Rehnsfeldt, 2008).
70
ABC semantics of human induced disaster corpus
For Natural disaster, the ABCs were combined for Human Induced Disaster. Figure 7.9 shows
the central themes of four disaster types: terrorist attack, fire, industrial and transport accidents.
Figure 7.9 ABC themes of human induced disaster class
In Figure 7.9, seven themes emerged from the analysis of ABC concepts, namely: Trauma,
Work, Escape, Screaming, Received, Feels and React. Further analysis of the concepts that
contributed to theme ‘Trauma’ (see Figure 7.10) revealed behavior concepts such as seeking,
express, adjustment; affective concepts such as loss, stress, and coping; and cognitive concepts
such as thoughts and beliefs.
The concept ‘Trauma’ is mapped to express (behavior), stress (affect), and thoughts
(cognition). In order to cope with the stressful situation, people expressed their thoughts, especially
in disasters such as fire. Studies have shown that in order to prepare people for making judgements
in hazard situations such as bushfire (Paton et al., 2006) and terrorist attack (Lasker, 2004), they
need to have (1) An intention to prepare and (2) An intention to seek information. The latter form
of intention requires people to express their needs for help, information and so forth.
71
Figure 7.10 ABC seed concepts of “trauma” theme for human induced disaster class
In Figure 7.10, the concept ‘risk’ is mapped to worry, mental (affect), perceive, alert, focus
(cognition) and decision making (behavior). There is a possibility that disaster situations such as
industrial accidents cause people to worry about the risk and they remain focused on the problem.
Miletti (1999) claimed that in order to mitigate industrial hazards, there is a need for people and
organizations to accept responsibility for the disaster so that losses may be curtailed through
change initiatives such as improved warning system, engineering design, building codes and
insurance coverage.
Attacks such as terrorist acts induce anxiety and people become emotional, sometimes angry
(affect), as they sense (cognition) the fears of those who experienced (behavior) the attacks. These
concepts mapped closely to each other in Figure 7.10.
In recent years, there have been several studies on terrorism (e.g. Sjoberg, 2005), but the
focus has been on the perpetrator rather than the victims. Jindapon and Neilson (2009) found that
citizens’ risk attitudes, such as anger expression, can affect the strategies of the terrorists and the
government. They claimed that a risk-averse population in a target country can induce terrorists to
commit more damaging, but less frequent, attacks.
72
ABC semantics of pandemic disaster class corpus
Figure 7.11 shows the central ABC themes of pandemic disaster. Eight themes emerged from
the analysis, namely: Behavior, Told, Woke, Passed, Broke, Saying, Complained and Wanting.
Figure 7.11 ABC central themes for pandemic disaster class
In Figure 7.12, ‘Behavior’ is the most dominant theme with concepts such as: action,
planning, change (behavior); arousal, emotions, panic (affect); risk, perception, and understanding
(cognition). The concept ‘behavior’ is closely mapped to risk and risk-taking and action (behavior),
spiritual (affect) and perception (cognition). The action includes educating themselves about the
risk.
73
Figure 7.12 ABC concepts of “behavior” theme for pandemic disaster class
Another concept that emerged from the analysis is ‘coping’, that is mapped to stigma and
emotional (affect), resulting in reappraisal (cognition) and adaptation (behavior). This situation is
true of HIV/AIDS and SARS patients who have to endure the stigma and discrimination of the
disease (Parker & Aggleton, 2003; Siu, 2008).
Disaster situation semantics of natural disaster class corpus
The corpora from personal stories (PS) and research articles (RA) were combined to evaluate
the central themes and seed concepts that emerged from the analyses. Figure 7.13 shows the
themes for natural disaster. Four central themes emerged from the natural disaster situation,
namely: Disaster, Earthquake, People and Water in the order of dominance.
In the ‘Disaster’ theme, as visualized in Figure 7.14, the most frequent concepts were
‘emergency’, ‘assistance’, ‘relief’, and ‘agencies’. This describes the situation as an ‘emergency’
where it required ‘assistance’ and ‘relief’. The ‘agencies’ in a natural disaster situation may be
represented by the people who provided aid and assistance to the victims and survivors.
74
Figure 7.13 Disaster situation themes for natural disaster class corpus
Figure 7.14 Disaster situation concepts of “disaster” theme for natural disaster class corpus
75
Disaster situation semantics of human induced disaster class corpus
In Figure 7.15, there were nine themes produced by the underlying concepts namely: Fire,
People, Smoke, Doorway, Morning, Sleeping, Called, Concern and Given. These themes described
the scenarios of the disaster rather than the attitudes of people.
Figure 7.15 Disaster situation themes of human induced disaster class corpus
In the ‘Fire’ theme (Figure 7.16), concepts such as ‘tunnel’, ‘attack’, ‘conditions’, ‘truck’ and
‘helicopter’ were prominent. The concept ‘tunnel’ was used widely to describe the conditions in
human induced disaster. In some incidents, the disaster happened in subways or tunnels. In others,
the rescuers had to build a tunnel to save the victims. Most of the incidents such as terrorist attack
also resulted in fire. Trucks in the map referred to fire engine trucks and helicopters to provide
support during the disaster.
76
Figure 7.16 “Fire” concepts of disaster situation for human induced disaster class corpus
Disaster situation semantics of pandemic disaster class corpus
Figure 7.17 shows the central themes of this disaster class. Nine themes were produced from
the analyses of the combined pandemic disaster class corpuses. These are: Factors, Health, Study,
People, Flu, Women, Global, Government, and Sick. The dominant theme is ‘Factors’. Given that
this concerns epidemic of diseases, it is not surprising for this theme to emerge. Figure 7.18 shows
the underlying seed concepts for theme ‘Factors’.
77
Figure 7.17 Disaster situation themes of pandemic disaster class corpus
Figure 7.18 “Factors” concepts of disaster situation for pandemic disaster class corpus
78
In Figure 7.18, the concepts derived from the analysis include ‘environmental’, ‘behavioral’,
‘influenced’, and ‘prevention’. ‘Environmental’ factors refer to the source that contributed to the
spread of diseases, including the surroundings. For example, in the case of swine flu and SARS,
the animal conditions that contributed to the epidemic. The spread of disease through human
populations across countries and worldwide turned the situation to become pandemic.
The ‘behavioral’ factor refers to people’s behavior such as sneezing, coughing, and so forth.
The pandemic scare may also influence their behavior, such as to be more careful of their
environment and to prevent from contracting the disease.
Conclusion
Natural and human-induced disasters affect everyone in their path. Some people are
nevertheless more vulnerable than others and suffer in different ways and to different extents. The
analysis of tsunami, for example, identified pre-existing factors that made people vulnerable, while
others became vulnerable as a result of disaster. Gender (e.g. women), age (e.g. children), personal
experience (e.g. loss, grief, fear) emerged as factors from the semantic analyses of disaster
experience text corpora.
In general, the maps are useful for understanding the nature of ABC and organizational
impact of services during disaster situations. The utility of the concept depends on its selection as a
leading concept in the dominant theme. There are many important concepts within the maps, but
suffice it to say that the maps represent the state of the scenario and that they summarize the
disaster data.
On the whole, the maps visualized the attitudes of people during different disaster situations,
and the kinds of services that could be rendered to manage the disaster. Since the articles that made
up the corpora vary in quality and quantity within and between disaster classes, the resulting map
is representative of the generic disaster class. To analyze specific disaster types, it would be
informative to scrutinize the semantic maps in Appendix D for ABC and Appendix E for disaster
situations. These maps are more detailed and explain each disaster type.
79
8
Hybrid Ontologies for Attitudinal Modeling
In this chapter we discuss the results of the study. The purpose of this study was to analyze ABCs
of disaster situations. Three classes of disasters were explored with four disaster types in each class.
The ABCs were derived from narratives and research articles of individual events. However, they
were analysed to construct a group attitudinal ontology.
ABCs, as modelled in Figure 2.1, are components of attitude. Understanding how people
behave, think and feel in the face of disaster provides an opportunity to forecast the behavior of
others who have yet to experience a disaster situation. The semantics derived from text mining
were mapped to the generic ontologies of attitude and disaster class in Chapter 4 to reveal the
concepts and requirements of people in disaster situations.
Hybrid Ontologies
Initially, ontologies are introduced as an explicit specification of a conceptualization (Gruber,
1993). Therefore, ontologies can be used in an integration task (e.g. developing a disaster
management system) to describe the semantics of the information sources and to make the contents
explicit. As such, they can be used for the identification and association of semantically
corresponding information concepts (Wache et al., 2001).
In the present research, we use a hybrid approach to integrate the attitudinal and
organizational (or disaster situational) ontologies to describe the semantics of people’s ABC and
disaster information framework for understanding the relationship between various concepts of
interest. A generic attitudinal and disaster situation ontology is given in Figure 4.1. Applying this
to the semantics derived from the Leximancer analysis, we can evaluate the ABC and the
requirements to prepare people for future risk-related situations.
80
Attitudinal ontologies
Natural disasters
Figure 8.1 represents the ontology for natural disaster based on combined data for tsunami,
earthquake, hurricane and flood. The semantics (seed concepts) were derived from the analyses of
the main theme, anticipation. The percentage values indicate the likelihood of mapping between
the linked concepts, and indicate the strength of the relationships that may be used in forecasting.
In anticipation of a natural disaster such as earthquake, hurricane and flood, people perceived
(cognition) the threat (risk perception) posed by the disaster situation and this induced a level of
fear (affect). The threat created concern (behavior) that may lead to action. The anticipation
dataset provides information about the level of concern, while the perception dataset influences
the perception level. The arousal dataset moderates the strength of emotion. The threat level is
determined by the anxiety dataset.
Figure 8.1 Attitudinal ontology for natural disaster
81
From Figure 8.1, the strength of the mapping relationship between perceived and threat
(19%) is relatively strong, followed by fear and threat (11%), then threat and concern (9%). This
suggests that cognition, in conjunction with affect, are at work during natural disasters, while
behavior occurs when the perceived threat is high. People tend to anticipate and demonstrate
concern as evidenced by 30% of concept mapping.
In other words, affect and cognition are equally important attitudinal components of natural
disasters, while behavior is present in all four disasters (see Chapter 7 – Figures 7.1 to 7.4), but the
degree of action is greater in tsunami than in the rest of the disasters. A study by Iemura,
Takahashi and Pradono (2006) found that survivors of tsunami held to floating objects, climbed
on house roofs, swam, climbed up trees, either intentionally or unintentionally. The survivors felt
that more could have survived if they ran immediately after the big earthquake. Since tsunami
occurs suddenly, the need to act is spontaneous; the element of fear sets in afterwards.
Human Induced Disasters
Figure 8.2 illustrates the ontology of 4 disaster types: terrorism, fire, industrial and transport
accidents. These disasters are different in character as described in Chapter 3. But in general, they
tend to induce greater reactions in people.
82
Figure 8.2 Attitudinal ontology for human induced disaster
In Figure 8.2, for the central theme trauma, it can be seen that trauma (affect) from the stress
dataset (26%) mapped strongly to vulnerability (risk perception) at 52%. People sense (cognition)
revealed from the belief dataset (10%) the risk which affected their vulnerability (17%). This
caused them to be desperate (behavior) which triggered action (33%) that is confirmed by the
dataset ran (33%).
In HID, it can be forecasted that people would act quicker than in a ND situation. This means
that behavior occurs before cognition, while affect in the form of fear or panic is the triggering
factor (Craske, 1999). This is because people feel vulnerable to the traumatic experience, and they
sense the approaching threat, as illustrated in Figure 8.3.
83
Figure 8.3 Relationship of worry, anxious apprehension and fear (or panic) (Craske, 1999).
Industrial accidents may also create the same reaction in victims. For example, a German
weekly, Der Spiegel, described the behavior of engineers and the fire department when electrical
cables caught fire in a Spanish nuclear power plant. Panic broke out in the control room, and the
engineers were all screaming at the same time and scrambling to get out. When the fire department
arrived, not a single engineer was willing to direct the efforts to extinguish the blaze (Fietkau,
1990).
Pandemic disasters
In Figure 8.4, the stigma (risk perception) associated with the disease caused victims to worry
(affect). This is evident from the emotion dataset (18%). The stigma is learnt from knowledge
(13%) of the disease. There is a tendency to perceive (cognition) the stigma which might lead to
confront (behavior) the situation through coping (25%) behavior.
With pandemic disasters, affect plays a greater role than cognition or behavior. Central to the
problem is the task of managing the negative emotions (e.g. shame, sadness, helplessness, fear,
worry) associated with serious illness and its consequences (Siegel & Schrimshaw, 2000). People
tend to worry more than taking action, especially in the case of HIV/AIDS. Depending on the
severity of the emotion induced by the risk, action may be slower than expected. This includes
seeking medical treatment (Finkenauer & Rimé, 1998).
84
Figure 8.4 Attitudinal ontology for pandemic disaster
Disaster situational ontologies
Natural disasters
Figure 8.5 illustrates the mapping of concepts in natural disasters. The emergency (disaster
situation) created by the disaster type requires assistance (situation need), which in turn requires
rescue aid (services) in the form of agencies (people). The type of rescue aid may be food (15%).
The agencies comprise of workers (10%), the type of assistance depends on the needs (12%), and
the relief effort (17%) will comply with the nature of the emergency situation.
85
Figure 8.5 Disaster situation ontology for natural disaster
This ontology represents the organizational aspects of disaster management in natural disaster
situations.
Human induced disasters
Figure 8.6 presents the concepts of HID. Due to safety (18%), a fire emergency (disaster
situation) would require the search of tunnels (situational need), depending on the risk (18%). The
tunnel requires the effort of rescue (services) workers (17%). The rescue services will work with
the users (people) in the tunnel who were identified by families (11%).
This scenario is typical of fire disasters and also for terrorism. People may be trapped in
tunnels (e.g. subways) due to fire or bombing (97%). Firemen also might be required to look into
tunnels or make tunnels as a way of controlling the outbreak of fire (31%). The relationship
between tunnel and rescue services is also moderately high, 26% (Associated Press, 2008; Lindley,
1997).
86
Figure 8.6 Disaster situation ontology for human induced disaster
Pandemic disasters.
Figure 8.7 presents the ontology for pandemic. An outbreak (disaster situation) caused by
infectious virus (26%) requires management (situational need), and this depends on the crisis
created by the outbreak (11%). The management requires a medical network (services) to provide
treatment (24%). The network may include government or medical staff (people) of hospitals
(11%) to manage the problem.
This scenario is typical of disease outbreak where the organization of teams requires
networking among countries and medical staff.
87
Figure 8.7 Disaster situation ontology for pandemic disaster
Summary
From the semantic framework it may be summarized that attitudinal ontologies are useful in
describing the ABC of people, while disaster situational ontologies are relevant in understanding
the organization of services that meet the needs of the disaster situation. For disaster management,
both attitudinal and organizational (disaster situation) ontologies are required, as they inform about
the attitudes of people in relation to their environment, see Figure 2.4. Most disaster management
systems focus on recovery plans without understanding people’s attitudes. Clearly, people’s affect,
behavior and cognition vary with the type of risk, and the response has to match their ABC
requirements.
88
9
Summary and Future Work
This chapter summarizes phase 1 activities. It also outlines the proposal for future work that takes
the present research as a point of departure.
SUMMARY OF FINDINGS
The goal of the project was to model risk attitude in the context of disaster experience.
Following Breckler (1984), attitude is made up of its tripartite components: Affect, Behavior and
Cognition (ABC). Affect is feeling, Behavior is doing and Cognition is thinking. A conceptual
attitudinal model was used to map the relationship of ABC components (see Figure 2.5, Chapter 2).
This relationship was further mapped in a semantic framework specified in Figure 4.1 (Chapter 4).
The semantics of risk perception, ABC was used to build several ontologies for attitudinal
modeling. In addition, organizational ontologies became necessary to map the relationship of
entities for disaster management. A hybrid ontology architecture that combined attitudinal and
organizational ontologies is shown in Figure 9.1 below.
Figure 9.1 maps the disaster situation ontology to attitudinal ontology at the risk perception
node. Understanding people’s risk attitudes alone is inadequate to support relief effort and vice
versa. For disaster management, it is important to identify the attributes of people and environment,
so that a comprehensive risk attitude model may be developed. This is to enable better forecasting
of group behavior.
Guiding the research activity is a methodology that specified the process for ABC
classification, data preparation and analysis (Chapter 5). To contextualize the model, a taxonomy
of disasters was provided in Chapter 3. In Chapter 6, we confirmed the semantic processing
capability of LSA and LXM. Given the greater usability of LXM and visualization of results in
concept maps, it was used for the semantic mining.
89
Figure 9.1 People-Environment Ontology Architecture
90
Chapter 7 investigated ABC using text corpora from personal stories and research articles.
Both generic and specific attitudinal and disaster situational ontologies were derived for natural
disasters, human induced disasters and pandemic disasters. Attitudinal ontology describes people’s
risk attitude, while disaster situational ontology describes the environment as illustrated in Figure
2.4 of the People-Environment systems model (Chapter 2).
Chapter 8 discussed the relevance of the semantics findings in relation to ABC. The mapping
of concepts at the node indicates the likelihood that these concepts are strongly linked. Overall,
there were more concepts for Behavior than for Affect and Cognition. To forecast socio-cultural
attitudes in disaster risk situations, Table 9.1 presents some expectations of ABC that may be
uncovered in future work.
Table 9.1. Attitudinal forecast as a function of disaster type, pre-warning and risk perception
Disaster type Pre-warning Risk Perception Expected ABC of people Natural Disasters Tsunami Sometimes no pre-
warning High risk Behavior – rushed, shouting, screaming, swim,
climbed, diving, crying, pulled, held, grabbed Affect – calm, lucky, loved, scared, happy Cognition – thinking, feel, decided, remember, forget
Earthquake None – but people are used to it
High-Moderate risk Behavior – breathing, avoidance, used, coping, planning, given Affect – relief, traumatic, fear, depression, suicidal Cognition – perceived, focus, reactions
Hurricane Yes - through mass media
Moderate risk Behavior – avoid, approach, seek, cope, helped, stress Affect – fear, traumatic, emotional, stress Cognition – self-efficacy, focus, perception, learn
Flood Yes - gradual build-up
Low-moderate risk Behavior – cope, work, provide, support, take, checking, control Affect – feel, threat, trauma, concern, depression, emotional Cognition – believe, sense, perception, adaptive
Human Induced Terrorism None High risk Behavior- provide, responded, choose, coping,
searching, making Affect – anxiety, anger, fear Cognition – attention, identify, beliefs thoughts
Fire None Moderate risk Behavior – evacuate, rescue, calling, pick, wait, drive Affect – happy, panic, fear, anxiety Cognition – decision- making, perceive, knowledge
Industrial Accidents
Immediate or gradual build up
Moderate risk Behavior- explain, avoid, obtain, working, gathered, observed Affect - danger, lost, fatalistic Cognition – perceived, attention, decisions
Transport accidents
None Moderate risk Behavior – driving, breathe, entered, looked, crying, working, ran Affect – distress, shock, fear Cognition – described, thought, realised
91
Pandemic Swine flu Relatively slow
build-up Low risk Behavior – breathing, avoid, sneezing, practice,
examined, using Affect – anxiety, distress, fear Cognition – thought, decision, emotions, ability
SARS Relatively slow build-up
Low risk Behavior – coughing, spitting, ordered, told, called, glanced, laying Affect – helpless, frightened, scary Cognition – thought, knows, experiencing
HIV/AIDS Ignore caution High risk Behavior – care, cope, confront, provides, reflects, stands, control Affect – anxiety, fear, anger, stress, arousal Cognition- decide, believe, emotion, perceived, attention
Tuberculosis Relatively slow build-up
Moderate risk Behavior – coughing, eating, share, working, travel, stay Affect – sick, suffering, afraid, sad Cognition – focus, thinking, known
FUTURE WORK
The next phase of activities involves gathering data from human subjects. This will be done
experimentally in the field. The aim is to measure people’s attitudes that can lead to behavioral
forecasting.
Research Design
The objectives of this study are:
1) To develop the attitudinal model for behavior forecasting. Data will be collected directly
from people in four cultural settings in Southeast Asia;
2) To validate the ABC concepts that was mined in Study 1 using: (a) leximancer analysis.
(b) ANOVA to compare the risk attitudes of cultural groups, and (c) correlation to assess
the relationship of concepts. The results will be used to analyze risk taking behavior for
various hazards in natural and human induced disasters.
The approach involves an experimental design using a risk attitudinal board, akin to the
citarasa elicitation system (Khalid et al., 2009) to measure risk perception, situational awareness
and cultural values. A control group will be used to compare the outcome of the risk evaluations.
Risk attitudinal board
Figure 9.2 provides a summary of the process used for assessing risk attitude.
92
Figure 9.2. Flow diagram of risk attitudinal board
Sort images according to
severity of risk from
lowest to highest risk
START
4
Images
END
Log In
Risk Perception
RegisterNew User
Existing User
Risk Assessment
Sudden >‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐< Delayed Common >‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐< Unusual Safe >‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐< Fatal
No damage >‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐< Much damage
Situation Awareness Assessment
Open‐endedWhat will happen next? How do you know?
Probe each answer.
Still
pictures
or video
Close‐
ended
Social Cultural Group Values Assessment
ABC: In this risk situation, what would you do first, and why? Concern: What are your major concerns relating to yourself in a risk situation, and why? Responsibility: In a risk situation do you feel responsible for others, and why? Assistance: How could other people help you, and why?
Open & close‐ended
93
The first part of the process in Figure 9.2 involves perception of risk through three steps:
1. Sort disaster images according to risk perceived,
2. Rank severity of risks from low to high risk, and
3. Rate the level of threat as perceived.
The second part of the process requires assessment of situational awareness of the disaster
scenarios. Participants will be shown a disaster situation to react to, and to predict what would
happen next. They will not be given responses to select. A control group will be provided with
answers most appropriate to the situation. This will make it possible to compare forecasting ability
as a function of the presence or absence of diagnostic information.
The final part of the evaluation process measures group values in relation to their ABC,
concerns, responsibility, and assistance required in disaster situations. Examples of questions:
ABC. In this risk situation, what would you do first, and why?
Concern. What are your major concerns relating to yourself in a risk situation, and why?
Responsibility. In a risk situation do you feel responsible for others, and why?
Assistance. How could other people help you, and why?
Both projective and why-why-why probe techniques will be used to elicit forecasting information
(Khalid et al., 2010a). An outline of the study is given in Appendix F.
In conclusion, we have demonstrated, in the present study, the importance of building
ontology for attitudinal modeling that identifies ABC semantics across heterogenous sources of
disaster information. The semantics may be applied to forecast risk attitudes of people in different
cultural settings so that a more comprehensive model of attitude may be developed for disaster
management.
94
REFERENCES Abelson, R. P., Kinder, D. R., Peters, M. D., and Fiske, S. T. (1982). Affective and semantic
components in political person perception. Journal of Personality and Social Psychology, 42, 619–30.
Abeyagunawardene, L. (2008). How I survived a plane crash. The Sunday Times, Dec 14, 2008. Adams, D.L. (Ed.). (1995). Health Issues For Women of Color: A Cultural Diversity
Perspective. Thousand Oaks: SAGE Publications. Aini, M.S. and Fakhrul’l-Razi, A. (2008). Management of inquiries into disasters: Experts’ views
and perspectives. Journal of Emergency Management, 6, 5, 37-50. Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl & J.
Beckman (Eds.), Action-control: From cognition to behavior (pp. 11- 39). Heidelberg, Germany: Springer. Retrieved November 1, 2007, from http://www.people.umass.edu/aizen/publications.html
Ajzen, I. (2001). Nature and operation of attitudes. Annual Review of Psychology, 52, 27–58. Allen, C. T., Machleit, K. A., and Kleine, S. S. (1992). A comparison of attitudes and emotions as
predictors of behavior at diverse levels of behavioral experience. Journal of Consumer Research, 18, 493–504.
Altheide, D.L. (2006). Terrorism and the politics of fear. Cultural Studies <=> Critical Methodologies, 6, 415.
Anderson, J. R. (1983). The Architecture of Cognition. Cambridge, MA: Harvard University Press. Anderson, J. R. (1996). ACT: A simple theory of complex cognition. American Psychologist, 51,
355-365. Asian Disaster Reduction Center (2008). Glossary on natural disasters, Available at:
http://www.adrc.or.jp/ Retrieved April 2008. Associated Press (2008). Fire in tunnel under English Channel suspends traffic; 6 hospitalized,
Thursday, September 11, 2008. Audi, R. (1995). The Cambridge Dictionary of Philosophy, Cambridge University Press,
Cambridge. Ballew, M. (2007). My personal experience in an earthquake in Paso Robles, California.
Associated Content, Retrieved May 3, 2010, from http://www.associatedcontent.com/article/921457/my_personal_experience_in_an_earthquake.html?cat=7
Banbury, S., and Tremblay, S. (2004). A Cognitive Approach to Situation Awareness: Theory, Measurement and Application. Aldershot, UK: Ashgate.
Barlow, D. H. (2000). Unraveling the mysteries of anxiety and its disorders from the perspective of emotion theory. American Psychologist, 55, 11, 1245-1263.
Baumeister, R.F., Vohs, K.D., SeWall, C.N. and Zhang, L. (2007). How emotion shapes behavior: feedback, anticipation, and reflection, rather than direct causation. Personal Social Psychology Review, 11(2), 167-203.
95
Ben-Zur, H., and Zeidner, M. (2009) Threat to life and risk-taking behaviors: a review of empirical findings and explanatory models. Personality and Social Psychology Review, 13, 2, 109-128.
Berger, P.L. (2002). The Cultural Dynamics of Globalization. In: P.L. Berger & S.P. Huntington (Eds.), Many Globalizations: Cultural Diversity in the Contemporary World. New York: Oxford University Press.
Berlyne, D.E. (1963). Motivational problems raised by exploratory and epistemic behavior, in Koch, S. (Ed.), Psychology: A Study of a Science, McGraw-Hill, New York, NY.
Block, M., and Seigel, R. (2006). Voices of Chernobyl: survivors' stories. National Public Radio Transcript, All Things Considered, April 21, 2006.
Bogen, K.T, and Jones, E.D. (2006). Risks of mortality and morbidity from worldwide terrorism: 1968–2004. Risk Analysis, 26, 45–59.
Bolin, R. (1985). Disaster characteristics and psychosocial impacts. In B. Sowder (Ed.), Disasters and Mental Health: Selected Contemporary Perspectives (pp. 3–28). Rockville, MD: National Institute of Mental Health.
Bonanno, G.A. (2004). Loss, trauma, and human resilience: have we underestimated the human capacity to thrive after extremely aversive events? American Psychologist, 59, 20-28.
Bonanno, G.A. (2005). Resilience in the Face of Potential Trauma, Current Directions in Psychological Science, 14, 3, 135-138.
Boon, H., Ruiter, R. A. C., James, S., van den Borne, B., Williams, E., and Reddy, P. (2010). Correlates of grief among older adults caring for children and grandchildren as a consequence of HIV and AIDS in South Africa, Journal of Ageing and Health, 22(1), 48-67.
Borum, R. (2004). Psychology of Terrorism. Tampa: University of South Florida. Breckler, S. J. (1984). Empirical validation of affect, behavior, and cognition as distinct
components of attitude. Journal of Personality and Social Psychology , 47 , 1191-1205. Breckler, S. J., and Wiggins, E. C. (1992). On defining attitude and attitude theory: Once more
with feeling. In A. R. Pratkanis, S. J. Breckler, & A. G. Greenwald (Eds.) Attitude Structure and Function (pp. 407-427). Hillsdale, NJ: Erlbaum.
Brown, B.R., and Granick, S. (1983). Cognitive and psychosocial differences between internal and external locus of control aged persons.
Buckle, P. (Ed.) (2004). Special issue: New perspectives on vulnerability. International Journal of Mass Emergencies and Disasters, 22, 5-109.
Buckner R.L., and Vincent J.L. (2007). Unrest at rest: Default activity and spontaneous network correlations. Neuroimage, 37:1091-6.
Buckner, R.L. and Carroll, D.C. (2007). Self-projection and the brain. Trends in Cognitive Science, 11: 49-57.
Carballo, M., Heal, B., and Horbaty, G. (2006). Impact of the tsunami on psychosocial health and well-being. International Review of Psychiatry, 18, 3, 217–223.
Card, S. (2003). Information Visualization. In: J.A. Jacko and A. Sears (Eds), The Human-Computer Interaction Handbook, Lawrence Erlbaum Associates, Mahwah, NJ, pp. 544-582.
Carver, C. S., Sutton, S. K., and Scheier, M. F. (2000). Action, emotion, and personality: Emerging conceptual integration. Personality and Social Psychology Bulletin, 26, 741-751.
Centre for Research on the Epidemiology of Disasters (2010). Retrieved 1 September 2010, http://www.cred.be/publications
Clore, G. L., and Huntsinger, J. R. (2007). How emotions inform judgment and regulate thought. Trends in Cognitive Science, 11, 393-399.
96
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20, 1, 37–46.
Cohen, J. (1968). Weighed kappa: Nominal scale agreement with provision for scaled disagreement or partial credit. Psychological Bulletin, 70, 213–220.
Craske, M. G. (1999). Anxiety disorders: Psychological approaches to theory and treatment. Denver, CO: Westview Press/Basic Books.
Cutter, S. L., Bryan J. B., and W. L. S. (2003). Social vulnerability to environmental hazards. Social Science Quarterly, 84, 1, 242-261.
Dalgard, O.S. and Haheim, L.L. (1998). Psychosocial risk factors and mortality: a prospective study with special focus on social support, social participation, and locus of control in Norway, Journal of Epidemiology and Community Health, 52, 476-81.
Damasio, A. R. (1995). Toward a neurobiology of emotion and feeling: operational concepts and hypotheses. The Neuroscientist, 1, 19-25.
Davis, M. (1992 a). The role of the amygdala in fear and anxiety. Annual Review of Neuroscience, 15, 353-375.
Dennis, C.D, and Mitterer, J.O. (2008). Introduction to Psychology: Gateways to Mind and Behavior. Cengage Learning. USA.
Derrico, K. (2009). Real life stories of adverse reactions to the H1N1 Swine Flu Vaccine. Retrieved Jan 11, 2010, from http://www.examiner.com/alternative-spirituality-in-atlanta/real-life-stories-of-adverse-reactions-to-the-h1n1-swine-flu-vaccine
Dillard, J. (1994). Rethinking the study of fear appeals: An emotional perspective. Communication Theory, 4, 295-323.
Dobbs, M. (2004). Paradise Lost. Reader's Digest, December 2004. Dolan, R.J. (2002). Emotion, cognition and behavior, Science, 298, 1191-1194. Dougall, A.L., Hayward, M.C., and Baum, A. (2005). Media exposure to bioterrorism: Stress and
the anthrax attacks. Psychiatry, 68, 28-42. Douglas, M. (1978). Cultural Bias. Occasional Paper no. 35, Royal Anthropological Institute of
Great Britain and Ireland. Dumais, S.T. (1994). Latent semantic indexing (LSI) and TREC-2. In: Harman D (Ed.)
Proceedings of The Second Text Retrieval Conference (TREC2), 105-116. Dunn, E. W., Brackett, M. A., Ashton-James, C., Schneiderman, E., and Salovey, P. (2007). On
emotionally intelligent time travel: Individual differences in affective forecasting ability. Personality and Social Psychology Bulletin, 33, 85-93.
DuPraw, M.E., and Axner, M. (1997). Working on common cross-cultural communication challenges. Retrieved 5 April 2006, http://www.wwcd.org/action/ampu/scg.html
Ebata, (1995). Geoinformation cartography, Geomatics solutions for disaster management, Duolunduo, Canada, pp. 185-199.
Edwards, K. (1990). The interplay of affect and cognition in attitude formation and change. Journal of Personality and Social Psychology, 59, 202–216.
Ellis, A. (2001). Overcoming Destructive Beliefs, Feelings, and Behaviors: New Directions for Rational Emotive Behavior Therapy. Amherst, New York: Promotheus Books.
Epstein, S. (1994). Integration of the cognitive and psychodynamic unconscious. American Psychologist, 49 (8), 709-724
Estes, T.H. (1999). Strategies for reading to learn: Semantic Maps, Retrieved 10 July 2010, http://www.readingquest.org/edis771/ semantic_maps.html
Federal Emergency Management Agency (2010), Type of disasters, Retrieved 1 September 2010, http://www.fema.gov/hazard/types.shtm.
97
Fietkau, H.J. (1990). Accident prevention and risk communication in environmental protection: a sociopsychological perspective. Organization Environment, 4, 4. 277-289.
Finkenauer, C. and Rimé, B. (1998). Keeping emotional memories secret: health and subjective well-being when emotions are not shared. J Health Psychology, 3, 47.
Finucane, M. L., Peters, E., and Slovic, P. (2003). Judgment and decision making: The dance of affect and reason. In S. L. Schneider & J. Shanteau (Eds.), Emerging Perspectives on Judgment and Decision Research. Cambridge, UK: Cambridge University Press.
Fischhoff, B., Slovic, P., Lichtenstein, S., Read, S., and Combs, B. (1978). How safe is safe enough? A psychometric study of attitudes toward technological risks and benefits. Policy Sciences, 9, 127-152.
Fiske, D. W., and Maddi, S. R. (1961). Functions of Varied Experience. Homewood, IL: The Dorsey Press.
Fleiss, J.L. (1971). Measuring nominal scale agreement among many raters, Psychological Bulletin, 76, 5, 378–382.
Fleiss, J.L. (1981). Statistical Methods for Rates and Proportions (2nd ed.). New York: John Wiley. ISBN 0471263702.
Foltz. P.W., Kintsch, W., and Landauer ,T.K. (1998). The measurement of textual coherence with latent semantic analysis. Discourse Process, 25, 2&3, 285-307.
Frijda, N.H. (1986). The Emotions. Cambridge ; New York : Cambridge University Press. Furby, L. and Beyth-Marom, R. (1992). Risk taking in adolescence: A decision making
perspective. Developmental Review, 12, 1-44. Galea, S., Nandi, A., and Vlahov, D. (2005). The epidemiology of posttraumatic stress disorder
after disasters. Epidemiologic Review, 27, 78-91. Gilbert, D. T., Gill, M. J., and Wilson, D. T. (2002). The future is now: Temporal correction in
affective forecasting. Organizational Behavior and Human Decision Processes, 88, 690-700. Gilovich, T., Griffin, D., and Kahneman, D. (Eds.) (2002). Heuristics and Biases: The Psychology
of Intuitive Judgment. New York: Cambridge University Press. Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of simple plans. American
Psychologist, 54, 493-503. Gonzalez, A., Solomon, S.E., Zvolensky, M.J. and Miller, C.T. (2009). The interaction of
mindful-based attention and awareness and disengagement coping with HIV/AIDS-related stigma in regard to concurrent anxiety and depressive symptoms among adults with HIV/AIDS, Journal of Health Psychology, 14, 403.
Grandien, C., Nord, L., & Strömbäck, J. (2005). Efter flodvågskatastrofen (after the Tsunami disaster). Krisberedningsmyndighetens temasserie: Stockholm.
Grandjean, D., Rendu, A-C., MacNamee, T. and Scherer, K.R. (2008). The wrath of the gods: appraising the meaning of disaster, Social Science Information, 47, 187.
Griffin, R.J., Yang, Z., Huurne, E.T., Boerner, F., Ortiz, S. and Dunwoody, S. (2008). After the Flood: Anger, Attribution, and the Seeking of Information, Science Communication, 29, 285.
Gruber, T. (1993). A translation approach to portable ontology specifications. Knowledge Acquisition, 5, 2, 199–220.
Gruber, T. (1995). Toward principles for the design of ontologies used for knowledge sharing. International Journal of Human-Computer Studies, 43, 5-6, 907-928.
Gyekye, S. A. (2003). Causal attributions of Ghanaian industrial workers for accident occurrence: Miners and non-miners perspective. Journal of Safety Research, 34, 5, 533-538.
Hoerger, M., Quirk, S.W., Lucas, R.E., and Carr, T.H. (2009). Immune neglect in affective forecasting. Journal of Research in Personality, 43, 91–94.
98
Hollnagel, E., D.D. Woods, and N. Leveson (2006). Resilience Engineering: Concepts and Precepts. Aldershot, UK: Ashgate.
International Federation of Red Cross and Red Crescent Societies (2005). World disaster reports, Retrieved 25 June 2008, www.ifrc.org/publicat/wdr2005/
Isen, A. M., Shalker, T. E., Clark, M., and Karp, L. (1978). Affect, accessibility of material in memory, and behavior: A cognitive loop? Journal of Personality and Social Psychology, 36, 1–12.
Jackson, D. (2007). 9-11-2001 And how I survived that day. Retrieved Nov 18, 2009, http://Blog.Scoop0901.Net/Life/How-My-Daughter-And-I-Survived-The-September-11-Terror-Attacks/
Jindapon, P. and Neilson, W.S. (2009). The impact of societal risk attitudes on terrorism and counterterrorism. Economics & Politics, 21, 3, 433-451
Johnson, E. J., and Tversky, A. (1984). Dimensional analysis of risk perceptions. Journal of Experimental Psychology (General), 113, 53-64.
Jung, C. G. (1971). Psychological types (Collected works of C. G. Jung, volume 6, Chapter X) Kahneman, D., and Tversky, A. (1979). Prospect theory: An analysis of decision under risk.
Econometrica, 47, 263–292. Kahneman, D., and Tversky, A. (Eds.). (2000). Choices, values and frames. New York:
Cambridge University Press. Khalid, H.M., Helander, M.G., Radha, J.K., Abd Karim, H.R. and Yang, X. (2008). Citarasa
Behavioral System for Sociocultural Analysis of Risk Attitudes toward Disasters/Hazards. AOARD/AFOSR Proposal, July 1, 2008.
Khalid, H.K., Abd Karim, H.R. and Radha, J.K. (2009). Citarasa System for elicitation and analysis of affective needs, Proceedings of IEA Congress 2009, Beijing, 9-14 August, 2009.
Khalid, H.M., Opperud, A., Radha, J.K., Xu, Q., and Helander,M.G. (2010a). Elicitation and analysis of citarasa for affective vehicle design, Theoretical Issues in Ergonomics Science, forthcoming.
Khalid, H.M., Radha, J.K., Yang, X. and Helander, M.G. (2010b). Analyzing ABC of disaster experience using text mining. In: D. Schmorrow and D. Nicholson (Eds), Advances in Cross-Cultural Decision Making. London: Taylor & Francis.
Klein, H.A., Lin, M.H., Radford, M., Masuda, T., Choi, I., Lien, Y., Yeh, Y.Y., and Chu, C. (2006). The Rosetta Project: Measuring National Differences. AOARD Report.
Kobes, M., Helsloot, I., De Vries, B., and Post, J. G. (2010). Building safety and human behaviour in fire: a literature review. Fire Safety Journal, 45, 1, 1-11.
Kreps, G.A. (2001). Disaster, Sociology of. In: N.J. Smelser & P. B. Bates (Eds.), International Encyclopedia of the Social and Behavioral Sciences. Amsterdam: Elsevier, pp. 3718-3721.
Lachman, M.E., and Weaver, S.L. (1998). The sense of control as a moderator of social class differences in health and well-being. Journal of Personality and Social Psychology, 74, 763-773.
Lan Lin, C., Chi Peng, Y., and Hung Tsai, C. (2004). Lessons learned from the anti-SARS quarantine experience in a hospital-based fever screening station in Taiwan. 7th East Asian Forum of Nursing Scholars Conference; March 19-20, 2004; Hong Kong, China.
Landauer, T.K, Foltz, P.W., and Laham, D. (1998). An introduction to Latent Semantic Analysis. Discourse Process, 25(2&3), 259-284.
Landauer, T.K, Laham, D., and Foltz, P.W. (2000). The intelligent essay assessor. IEEE Intelligent Systems and Their Applications, 15, 5, 27-31.
99
Landauer, T.K. and Dumais, S.T. (1997). A solution to Plato's problem: The Latent Semantic Analysis Theory of acquisition, induction, and representation of knowledge. Psychological Review, 104, 211-240.
Lang, P. J., Bradley, M. M., and Cuthbert, B. N. (1990). Emotion, attention, and the startle reflex. Psychological Review, 97, 377-395.
Lasker, R.D. (2004), Redefining Readiness: Terrorism Planning through the Eyes of the Public. New York, NY: The New York Academy of Medicine.
Latent Semantic Analysis (2008). Retrieved 1 June 2008, http://lsa.colorado.edu/ Lavine, H., Thomsen, C. J., and Zanna, M. P. (1998). On the primacy of affect in the determination
of attitudes and behavior: The moderating role of affective-cognitive ambivalence. Journal of Experimental Social Psychology, 34, 398–421.
LeDoux, J. E. (1995). Emotion: clues from the brain. Annual Review of Psychology, 46: 209-235. LeDoux, J.E. (2000). Emotion circuits in the brain. Annual Review Neuroscience, 23,155-184. LeDoux, J.E. (2003). Synaptic Self: How Our Brains Become Who We Are. New York: Penguin
Putnam. Lee, K. (2001). Hometown tornado touches everyone. Victims and the Media Program Newsletter,
Spring 2001. Lerner, J. S., and Gonzalez, R. M. (2005). Forecasting one's future based on fleeting subjective
experiences. Personality and Social Psychology Bulletin, 31, 454–466. Lerner, J. S., Gonzalez, R. M., Small, D. A., and Fischhoff, B. (2003). Effects of fear and anger on
perceived risks of terrorism: A national field experiment. Psychological Science, 14, 144–150.
Leximancer (2009). Leximancer manual. Retrieved 1 August 2009, http://www.leximancer.com/ Lichtenstein, S. and Slovic, P. (2006). The Construction of Preference. New York: Cambridge
University Press. Lichtenstein, S., Slovic, P., Fischhoff, B., Layman, M. and Combs, B. (1978). Judged frequency of
lethal events. J. Experimental Psychology Learning, 4, 551-578. Lindley, J. (1997). The Summit Tunnel incident, Loss Prevention Bulletin, 134, 14–19. Livanou, M., Kasvikis, Y., Basoglu, M., Mytskidou, P., Sotiropoulou, V., Spanea, E., Mitsopoulou,
T., and Voutsa, N. (2004). Earthquake-related psychological distress and associated factors 4 years after the Parnitha earthquake in Greece. Journal of European Psychiatry, March 2005, 20, 2, 137-144.
Loewenstein, G. F., Weber, E. U., Hsee, C. K., and Welch, N. (2001). Risk as feelings. Psychological Bulletin, 127, 267–286.
Lollar, K. (2010). The liminal experience: loss of extended self after the fire Qualitative Inquiry,16, 262.
Mann, H. (2007). Upplevd nyhetsrapportering och associerade känslor i samband med katastrofer: En explorativ studie. [News coverage and experienced feelings in connection to disasters: and explorative study]. Reports from Stockholm University. Retrieved 19 August 2010, http://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-6777.
Mano, H. (1991). The structure and intensity of emotional experiences. Multivariate Behavioral Research, 26, 389–411.
McClelland, S. (2003). SARS victims tell their stories. The Canadian Encyclopedia, Retrieved February 12, 2010 http://www.canadianencyclopedia.ca/ index.cfm?PgNm=TCE&Params=M1ARTM0012476
McIntyre, B. (2009). 25 years later I'm still here. Retrieved 20 April, 2010, http://www.PositivelyPositive.ca
100
Mileti, D.S. (1999), Disasters by Design: A Reassessment of Natural Hazards in the United States. Washington, DC: Joseph Henry.
Mileti, D.S. and Gailus, J.L. (2005).Sustainable development and hazards mitigation in the United States: Disasters by Design revisited, Mitigation and Adaptation Strategies for Global Change, 10 2, 491-504.
Miller, G. (2005). The tsunami’s psychological aftermath, Science, 309, 5737, 1030-3. National Research Council (1998). Modeling Human and Organizational Behavior. Washington,
DC.: The National Academies Press. National Research Council (2001). Grand Challenges in Environmental Sciences. Washington,
DC.: The National Academies Press. National Research Council (2006). Facing Hazards and Disasters. Washington, DC.: The National
Academies Press. Nisbett, R. E., and Ross, L. D. (1980). Human Inference: Strategies and Shortcomings of Social
Judgment. Englewood Cliffs, NJ: Prentice-Hall. Norman,D.A., Ortony, A, and Russell, D.M. (2003). Affect and machine design: Lessons for the
development of autonomous machines. IBM Systems Journal, 42, 1, 38-44. Norris, F. H., Friedman, M. J., Watson, P. J., Byrne, C. M., Diaz, E., and Kaniasty, K. (2002).
60,000 disaster victims speak: Part I. An empirical review of the empirical literature, 1981–2001. Psychiatry, 65, 207-239.
Nyamathi, A., Berg, J., Jones, T. and Leake, B. (2005). Predictors of perceived health status of tuberculosis-infected homeless. West J Nurs Res, 27, 896.
Ocaya, B. (2010). I didn’t want to live neither did I want to die, I just didn’t want to be there. Retrieved 22 May 2010, http://www.tbsurvivalproject.org/stories/Benjamin.html
Oeveren, K. V. (1997). Flood- Disaster 1953. Retrieved 17 May 2010, http://People.Zeelandnet.Nl/Voeveren/Verhaale.Htm
Öhman, A. (2005). The role of the amygdala in human fear: Automatic detection of threat, Psychoneuroendocrinology, 30, 10, 953-958.
Oltedal, S., Moen, B., Klempe, H., and Rundmo, T. (2004). Explaining Risk Perception. An evaluation of cultural theory. Norway: Rotunde Publication. Retrieved 17 June 2008, http://www.forskningsradet.no/servlet/ Satellite
Orasanu, J. and Connolly, T. (1993). The reinvention of decision making. In: G.A. Klein, J. Orasanu, R. Calderwood, R., and C.E. Zsambok (Eds.), Decision Making in Action: Models and Methods. Norwood, NJ: Ablex Publishing Corporation, p. 7.
Osberg, T. M. and Shrauger, J.S. (1986). Self-prediction: Exploring the parameters of accuracy, Journal of Personality and Social Psychology, 51, 1044-1057.
Ozer, E.J., Best, S.R., Lipsey. T.L., and Weiss, D.S. (2003), Predictors of posttraumatic stress disorder and symptoms in adults: A Meta-Analysis. Psychol Bull., 129, 1, 52-73.
Parker, R. and Aggleton, P. (2003). HIV and AIDS-related stigma and discrimination: a conceptual framework and implications for action. Social Science & Medicine, 57, 1, 13-24
Paton, D., Kelly, G., Burgelt, P.T., and Doherty, M. (2006). Preparing for bushfires: understanding intentions. Journal of Disaster Prevention and Management, 15, 4, 566-575.
Pennebaker, T. (2009). Real time swine flu coverage: Stanford health experts regularly weigh in. Center For Health Policy, Stanford University. Retrieved 25 November 2009, http://healthpolicy.stanford.edu/news/swine_flu/
Peters, E. (2006). The functions of affect in the construction of preferences. In S. Lichtenstein & P. Slovic (Eds.). The Construction of Preference, New York: Cambridge University Press.
101
Plapp, T. and Werner, U. (2006). Understanding risk perception from natural hazards: examples from Germany. RISK 21 – Coping with Risks due to Natural Hazards in the 21st Century. p. 101-108.
Poltzer, P. (2008). Tuberculosis: A new pandemic. CNN News, Retrieved 10 May 2010, http://articles.cnn.com/2008-11-17/health/tb.pandemic_1_xdr-tb-anti-tb-drugs-tuberculosis?_s=PM:HEALTH
Quarantelli, E.L. (1998). What is a Disaster? Perspectives on the Question. New York: Routledge. Quinn, J. (2009). New York plane crash: Hero pilot did 'absolutely the right thing'. Telegraph
News, Jan 16. Raholm M B, Arman M and Rehnsfeldt A. (2008). The immediate lived experience of the 2004
tsunami disaster by Swedish tourists. Journal of Advanced Nursing. 63, 6, 597-606. Raviv, A., Sadeh, A., Raviv, A., Silberstein, O., and Diver, O. (2000). Young Israelis’ reactions to
national trauma: The Rabin assassination and terror attacks. Political Psychology, 21, 299-322.
Reich, J.W. (2006). Three psychological principles of resilience in natural disasters. Disaster Prevention and Management, 15, 5, 793-798.
Reiman, T., and Oedewald, P. (2007). Assessment of complex sociotechnical systems - theoretical issues concerning the use of organizational culture and organizational core task concepts. Safety Science, 45, 7, 745-768.
Reis, D. L., and Gray, J. R. (2009). Affect and action control. In E. Morsella, J. A. Bargh, and P. Gollwitzer (Eds), The Oxford Handbook of Human Action. New York: Oxford Univ. Press.
Ripley, A. (2008). The Unthinkable: Who Survives When Disaster Strikes - and Why. New York: Crown Archetype Publishing.
Rodin, J., Timko, C. and Harris, S. (1985), The construct of control: biological and psychosocial correlates, in Lawton, M.P. and Maddox, G.L. (Eds), Annual Review of Gerontology and Geriatrics, 5. New York, NY: Springer.
Rosa, E. A. (2003). The logical structure of the social amplification of risk framework (SARF): Metatheoretical foundation and policy implications. In N. K. Pidgeon, R.E.and Slovic, P (Ed.), The Social Amplification of Risk (pp. 47-79). Cambridge: Cambridge University Press.
Rosenberg, M.J. and Hovland, C.I. (1960) Cognitive, affective and behavioral components of attitudes. In: Hovland, C.I. and Rosenberg, M.J. (Ed.), Attitude Organization and Change (pp.1-14), New Haven, CT: Yale University Press.
Ruby, C. L. (2002). The definition of terrorism. Analyses of Social Issues and Public Policy, 9–14. Russell, J.A. (2003). Core affect and the psychological construction of emotion, Psychological
Bulletin, 110, 145–72. Scarabis, M., Florack, A., and Gosejohann, S. (2006). When consumers follow their feelings: The
impact of affective or cognitive focus on the basis of consumers` choice. Psychology & Marketing, 23, 1015–1034.
Scherer, K. R., Shorr, A., and Johnstone, T. (Ed.). (2001). Appraisal Processes in Emotion: Theory, Methods, Research. Canary, NC: Oxford University Press.
Schneid, T.D. and Collins, L. (2001). Disaster Management and Preparedness. Lynden, WA: Lewis Publishers.
Schneirla, T.C. (1959). An evolutionary and developmental theory of bi-phasic processes underlying approach and withdrawal. In: M.R. Jones (Ed.) Nebraska Symposium on Motivation. Lincoln: University of Nebraska Press.
Schwarz, N. (2000). Emotion, cognition, and decision making. Cognition and Emotion, 14, 433–440.
102
Seery, M. D., Silver, R. C., Holman, E. A., Ence, W. A., and Chu, T. Q. (2008). Expressing thoughts and feelings following a collective trauma: immediate responses to 9/11 predict negative outcomes in a national sample. Journal of Consulting and Clinical Psychology, 76, 4, 657-667.
Shaluf, I.M. (2007). An overview on disasters. Disaster Prevention and Management,16, 5, 687 – 703.
Shavelson, R.J. and Stanton, G.C. (1975). Construct validation: methodology and application to three measures of cognitive structure, Journal of Educational Measurement, 12, 2, 67-85.
Sheth, A. (1999). Changing focus on interoperability in information systems: from system, syntax, structure to semantics. In: M.F. Goodchild, M. Egenhofer, R. Fegeas and C.A. Kottman (Eds), Interoperating Geographic Information Systems, Dordrecht, Netherlands: Kluwer Academic, pp. 530.
Shiloh, S., Guvenc, G., and Onkal, D. (2007). Cognitive and emotional representations of terror attacks: A cross-cultural exploration. Risk Analysis, 27(2), 397-409.
Siegel, K. and Schrimshaw, E.W. (2000). Gay/bisexual men coping with negative emotions: the cognitive strategies of HIV-infected. Journal of Health Psychology, 5, 517.
Siu, J.Y. (2008). The SARS-associated stigma of SARS victims in the post-SARS era of Hong Kong, Qualitative Health Research, 18; 729.
Sjöberg, L. (1996). A discussion of the limitations of the psychometric and cultural theory approaches to risk perception. Radiation Protection Dosimetry, 68, 219-225.
Sjöberg, L., (1979). Strength of belief and risk, Policy Sciences, 11, 39–57. Sjoberg, L., (2005). The perceived risk of terrorism. Risk Management: An International Journal 7,
43–61. Sjöberg, L., Moen, B.E. and Rundmo, T. (2004). Explaining Risk Perception (ISBN 82-7892-026-
5) Slovic, P. (1987). Perception of risk. Science, 236, 280-285. Slovic, P., Finucane, M. L., Peters, E., & MacGregor, D. G., (2004). Risk as analysis and risk as
feelings: some thoughts about affect, reason, risk, and rationality. Risk Analysis, 24, 2, 1-12. Slovic, P., Fischhoff, B., and Lichtenstein, S. (1985). Charactering perceived risk. In R. Kates, C.
Hohenemser & R. Kasperon (Eds.), Perilous progress: Managing the hazards of technology (pp. 99-125). Boulder, CO: Westview.
Smith, A.E. (2005). Leximancer Manual: Version 2.2. University of Queensland, Australia. Retrieved 1 September 2009, https://www.leximancer.com/
Smith, B. (2008). A look at terrorist behavior: How they prepare, where they strike. National Institute of Justice, Issue, 260.
Smith, C. A., and Ellsworth, P. C. (1985). Patterns of cognitive appraisal in emotion. Journal of Personality and Social Psychology, 48, 813–838.
Sojka, J. Z., and Giese, J. L. (1997). Thinking and/or feeling: An examination of interaction between processing styles. Advances in Consumer Research, 24, 438–442.
Solomon, M. R. (1996). Consumer Behaviour, Englewood Cliffs, NJ: Prentice Hall. Somasundaram, D., F.H., Asukai, N. and. Srinivasa Murthy, R. (2003). Natural and technological
disasters In: B.L. Green (Ed.), Trauma Interventions in War and Peace: prevention, practice, and policy, New York: Kluwer Academic, pp. 291-318.
SPSS (2009). SPSS, Retrieved 10 February 2009, http://www.spss.com/ Studer, R., Benjamins, V.R, and Dieter Fensel, D. (1998). Knowledge Engineering: Principles and
methods, Data & Knowledge Engineering, 25, 161-197.
103
Taatgen, N.A. & Anderson, J.R. (2008). ACT-R. In R. Sun (Ed.), Constraints in Cognitive Architectures. New York: Cambridge University Press, pp 170-185.
Takao, K., Motoyoshi, T., Sato, T., Fukuzondo, T., Seo, K., and Ikeda, S. (2004). Factors determining residents' preparedness for floods in modern megalopolises: the case of the Tokai flood disaster in Japan. Journal of Risk Research, 7, 7 & 8, 775 – 787.
Thompson, J. (2005). Surviving more than the fire. Retrieved 13 February 2010, http://www.Burnsurvivorsttw.Org/Stories/Jacq.Html
Tierney, K.J., Lindell, M., and Perry, R. (2001). Facing the unexpected: disaster preparedness and response in the United States. Washington, DC: Joseph Henry Press.
Toft, B. and Reynolds, S. (1994). Learning from Disasters. Oxford: Butterworth-Heinemann. Trope, Y., and Liberman, N. (2010). Construal level theory of psychological distance.
Psychological Review, 117, 440-463. Uschold, M. and Grüninger, M.(1996), Ontologies: Principles, Methods, and Applications,
Knowledge Engineering Review, 1, 96-137. Västfjäll, P. and Slovic, P. (2008). Affect, risk perception and future optimism after the tsunami
disaster. Judgment and Decision Making, 3(1), 64–72. Vicente M. and Simón, V. M. (1998). Emotional participation in decision- making, Psychology in
Spain, 2, 1, 100-107. Wache, H., Vogele, T., Visser, U., Stuckenschmidt, H., Schuster, G. Neumann, H. and Hubner, S.
(2001). Ontology-based integration of information — a survey of existing approaches. Proceedings of the International Workshop on Ontologies and Information Sharing, pp. 108-117 (August 2001).
Weems, C. F., Watts, S. E., Marsee, M. A., Taylor, L. K., Costa, N. M., Cannon, M. F., Carrion, V. G., and Pina, A. A. (2007). The psychosocial impact of hurricane Katrina: contextual differences in psychological symptoms, social support, and discrimination. Journal Of Behaviour Research And Therapy, 45, 10, 2295-2306.
Weinstein, N.D. (1989). Optimistic biases about personal risks, Science, 246, 4935, 1232 – 1233.
Wickens, C. D., and Hollands, J. G. (2000). Navigation and interaction in real and virtual environment. In Engineering Psychology and Human Performance (pp. 158-195). Upper Saddle River, NJ: Prentice Hall.
Wilson, T.D. and LaFleur, S.J. (1995). Knowing what you’ll do: effects of analyzing reasons on self-prediction. Journal of Personality and Social Psychology, 68, 21–35.
Wirga, M., and Bernardi, M. D., (2008). The ABCs of cognition, emotion, and action. Retrieved 24 June 2008 http://www.arcobem.com/publications/ABC%20of%20Cognition.html
World Health Organization (2010), http://www.who.int/whr/en/ Xu, W. and Zlatanova, S. (2007). Ontologies for disaster management response. In: Li, Zlatanova
& Fabbri (Eds.) Geomatics Solutons for Disaster Management, Lecture Notes in Geoinformation and Cartography, Berlin, Heidelberg: Springer-Verlag, pp. 185-200
Yamokoski, C.A. (2006). ABC’S of suicidology: the role of affect in suicidal behaviors and cognitions, Ohio University PhD Thesis. Retrieved 15 December 2009, http://drc.ohiolink.edu/handle/2374.OX/3814.
Zacharias, G.L., MacMillan, J., and Van Hemel S.B. (Eds.) (2008). Committee on Organizational Modeling: From Individuals to Societies, National Research Council.
Zsambok, C.E., and Gary A. Klein, G.A. (1997). Naturalistic Decision Making. Mahwah, NJ: Lawrence Erlbaum.